Bug 203373 – [ iOS Release ]: Layout Test http/tests/navigation/page-cache-domcache-pending-promise.html is a Flaky Timeout

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Russell Epstein]

The following layout test is a flaky timeout on iOS Release:

http/tests/navigation/page-cache-domcache-pending-promise.html

Steps to Reproduce:

run-webkit-tests --ios-simulator --no-retry --no-sample-on-timeout --child-processes 1 -g --iter 100 http/tests/navigation/page-cache-domcache-pending-promise.html

Test History:

https://results.webkit.org/?suite=layout-tests&test=http%2Ftests%2Fnavigation%2Fpage-cache-domcache-pending-promise.html&platform=ios

Diff:
--- /Volumes/Data/slave/ios-simulator-13-release-tests-wk2/build/layout-test-results/http/tests/navigation/page-cache-domcache-pending-promise-expected.txt
+++ /Volumes/Data/slave/ios-simulator-13-release-tests-wk2/build/layout-test-results/http/tests/navigation/page-cache-domcache-pending-promise-actual.txt
@@ -1,15 +1,5 @@
-Tests that a page with pending DOMCache activity goes into the page cache.
+#PID UNRESPONSIVE - WebKitTestRunnerApp (pid 92071)
+FAIL: Timed out waiting for notifyDone to be called
 
-On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
-
-
-pageshow - not from cache
-pagehide - entering cache
-pageshow - from cache
-PASS Page was restored from Page Cache
-PASS Cache.add() succeeded
-PASS !!restoredFromPageCache is true
-PASS successfullyParsed is true
-
-TEST COMPLETE
-
+#EOF
+#EOF

[Radar WebKit Bug Importer]

<rdar://problem/56590038>

[Chris Dumez]

Looking into this now.

[Chris Dumez]

Unfortunately, run-webkit-tests --ios-simulator --no-retry --no-sample-on-timeout --child-processes 1 -g --iter 100 http/tests/navigation/page-cache-domcache-pending-promise.htm

does NOT reproduce it for me :/

[Alexey Proskuryakov]

This test's run time is all over the place - 1 second, 6 seconds, 26 seconds, timeout...

[Chris Dumez]

(In reply to Chris Dumez from comment #3)
> Unfortunately, run-webkit-tests --ios-simulator --no-retry
> --no-sample-on-timeout --child-processes 1 -g --iter 100
> http/tests/navigation/page-cache-domcache-pending-promise.htm
> 
> does NOT reproduce it for me :/

No luck reproducing this on my second machine either.

[Chris Dumez]

Still not able to reproduce this.

[Chris Dumez]

I see this logging in stderr on the bots:
CONSOLE MESSAGE: line 23: Fetch API cannot load http://127.0.0.1:8000/navigation/resources/blank.txt due to access control checks.

But I think that's expected.

[Chris Dumez]

Created attachment 383652 [details]
Patch

[Chris Dumez]

Created attachment 383748 [details]
Patch

[Alexey Proskuryakov]

Comment on attachment 383748 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=383748&action=review

> LayoutTests/ChangeLog:12
> +        Speculative fixes:

This is failing EWS

[Chris Dumez]

Comment on attachment 383748 [details]
Patch

Will investigate.

[Chris Dumez]

(In reply to Chris Dumez from comment #11)
> Comment on attachment 383748 [details]
> Patch
> 
> Will investigate.

Very odd, this test is passing for me locally on Mac-wk2.

[Chris Dumez]

Created attachment 383804 [details]
Patch

[Chris Dumez]

I cannot reproduce it locally but it looks like caches.open() flakily times out on the iOS wk2 and macOS wk2 bots for some reason.

[Chris Dumez]

Created attachment 383878 [details]
Patch

[Chris Dumez]

Ping review? EWS was able to reproduce the flakiness previously so it looks like this did the trick.

[Aakash Jain]

This is slowing down EWS sometimes.

e.g.: In https://ews-build.webkit.org/#/builders/24/builds/5353 this test failed while running layout-tests, and EWS had to re-run layout-tests in which it passed. If the test wasn't flaky, the re-run wouldn't have been required.

[WebKit Commit Bot]

Comment on attachment 383878 [details]
Patch

Clearing flags on attachment: 383878

Committed r252769: <https://trac.webkit.org/changeset/252769>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.