The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7c3a826500167cf33bd0b2421fa310df91396d09

commit 7c3a826500167cf33bd0b2421fa310df91396d09
Author: johnme <johnme@chromium.org>
Date: Thu Aug 25 13:40:27 2016

Mark http/tests/serviceworker/foreign-fetch-cors.html flaky on Win

Failed on the following builds, amongst others:
7052 7040 7035 7031 7024 7023 6993 6988 6974
Bot: https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Win7%20%28dbg%29?numbuilds=200

BUG= 640947 
TBR=mek@chromium.org
NOTRY=true
NOTREECHECKS=true

Review-Url: https://codereview.chromium.org/2278133002
Cr-Commit-Position: refs/heads/master@{#414418}

[modify] https://crrev.com/7c3a826500167cf33bd0b2421fa310df91396d09/third_party/WebKit/LayoutTests/TestExpectations

They all just seem to tune out, so I suspect this might just be a combination of too many promise_tests in a single file, and testharness.js not taking into account that it serialized such tests when calculating timeouts...

Filed https://github.com/w3c/testharness.js/issues/208 for the testharness.js bug that's causing this.

It looks like a fix was landed at https://github.com/w3c/testharness.js/issues/208. Did we pull that change in?

I managed to pull that change into WPT last night. Today I was trying to pull in a newer version of WPT to get the testharness.js test in our tree, but the previous WPT roll ended up getting reverted instead. The next time WPT gets rolled we should get that change too...

As I think mek is away I'll take this.

[Blink-Worker Bug Triage] P1->P2 based on c#3

 Issue 658314  has been merged into this issue.

The virtual/mojo-service-worker variant is also flaky on Windows:

virtual/mojo-service-worker/http/tests/serviceworker/foreign-fetch-cors.html

See https://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=virtual%2Fmojo-service-worker%2Fhttp%2Ftests%2Fserviceworker%2Fforeign-fetch-cors.html

In addition, despite automatic retries it sometimes turns the build red: https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Win7%20(dbg)/builds/7814

When it fails, the output is:

CONSOLE ERROR: line 106: Refused to get unsafe header "X-ServiceWorker-ServerHeader"
CONSOLE ERROR: line 106: Refused to get unsafe header "X-ServiceWorker-ServerHeader"
CONSOLE ERROR: line 106: Refused to get unsafe header "X-ServiceWorker-ServerHeader"
CONSOLE ERROR: line 106: Refused to get unsafe header "X-ServiceWorker-ServerHeader"
This is a testharness.js-based test.
Harness Error. harness_status.status = 2 , harness_status.message = null
PASS Same origin fetch without CORS headers, not exposed 
PASS Same origin fetch without CORS headers, only origin exposed 
PASS Same origin fetch without CORS headers, headers and origin exposed 
PASS Same origin fetch without CORS headers, exposed to wrong origin 
PASS Same origin fetch with CORS headers, not exposed 
PASS Same origin fetch with CORS headers, only origin exposed 
PASS Same origin fetch with CORS headers, headers and origin exposed 
PASS Same origin fetch with CORS headers, exposed to wrong origin 
PASS Cross origin fetch without CORS headers, not exposed 
PASS Cross origin fetch with ACEHeaders header, not exposed 
PASS Cross origin fetch with ACEHeaders header, only origin exposed 
PASS Cross origin fetch with ACEHeaders header, headers and origin exposed 
PASS Cross origin fetch with ACEHeaders header, exposed to wrong origin 
PASS Cross origin fetch without ACEHeaders header, not exposed 
PASS Cross origin fetch without ACEHeaders header, only origin exposed 
TIMEOUT Cross origin fetch without ACEHeaders header, headers and origin exposed Test timed out
Harness: the test ran to completion.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/02c1ced8a1c950f6c08f8b0c61d589b482f706e1

commit 02c1ced8a1c950f6c08f8b0c61d589b482f706e1
Author: johnme <johnme@chromium.org>
Date: Fri Oct 21 18:04:31 2016

Mark virtual/mojo-service-worker/.../foreign-fetch-cors.html flaky

BUG= 640947 
TBR=falken@chromium.org
NOTRY=true
NOTREECHECKS=true

Review-Url: https://chromiumcodereview.appspot.com/2442043002
Cr-Commit-Position: refs/heads/master@{#426839}

[modify] https://crrev.com/02c1ced8a1c950f6c08f8b0c61d589b482f706e1/third_party/WebKit/LayoutTests/TestExpectations

 Issue 664144  is about a similarly named test. Not sure if it is a duplicate or not, please feel free to merge.

 Issue 664144  has been merged into this issue.

Apparently the testharness.js change that I hoped would fix the timeout issue didn't actually fix it... I'll investigate more to see what is going on here...

This really seems like somehow testharness.js that is being used by the WebKit Win7 (dbg) test bots is not the version that is checked in...
At least from the access logs for the failed test runs (after testharness.js got fixed) it is clear that none of the individual tests come anywhere close to the 10 seconds per individual test timeout testharness.js uses, yet the last (or sometimes the one before the last test) still is considered timed out after the overall test run has lasted 10 seconds. I guess for now I can just manually increase the per-test timeout to be more than 10 seconds, but with the current testharness.js that really shouldn't be necessary.

Actually, never mind, this is of course just the global testharness.js timeout that hits, not any per test timeout. So just another instance of #664268. 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/bdd208fae216332c37dfc202d77cf04cd8a1fbd6

commit bdd208fae216332c37dfc202d77cf04cd8a1fbd6
Author: mek <mek@chromium.org>
Date: Wed Nov 16 21:20:18 2016

Re-enable foreign-fetch-cors layout test.

With recent fixes in testharness.js and testharnessreport.js this test should
no longer be failing. Maybe this test should be added to SlowTests instead
but for now it seems to be fast enough to not get too close to the actual test
timeout.

BUG= 640947 

Review-Url: https://codereview.chromium.org/2511473002
Cr-Commit-Position: refs/heads/master@{#432629}

[modify] https://crrev.com/bdd208fae216332c37dfc202d77cf04cd8a1fbd6/third_party/WebKit/LayoutTests/TestExpectations