Some r_and_r.top_25 stories time out due to slow load. |
|||||||||
Issue descriptionrasterize_and_record_micro.top_25 failing on chromium.perf/Android One Perf Builders failed on: - Android One Perf: https://build.chromium.org/p/chromium.perf/builders/Android%20One%20Perf Seems to always be on espn.html Traceback (most recent call last): File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/story_runner.py", line 105, in _RunStoryAndProcessErrorIfNeeded state.RunStory(results) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/page/shared_page_state.py", line 332, in RunStory self._current_page.Run(self) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/page/__init__.py", line 114, in Run shared_state.page_test.RunNavigateSteps(self, current_tab) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/page/legacy_page_test.py", line 152, in RunNavigateSteps page.RunNavigateSteps(action_runner) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/page/__init__.py", line 123, in RunNavigateSteps url, script_to_evaluate_on_commit=self.script_to_evaluate_on_commit) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/actions/action_runner.py", line 187, in Navigate timeout_in_seconds=timeout_in_seconds)) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/actions/action_runner.py", line 63, in _RunAction action.RunAction(self._tab) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/actions/navigate.py", line 27, in RunAction tab.WaitForDocumentReadyStateToBeInteractiveOrBetter(time_left_in_seconds) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/browser/web_contents.py", line 103, in WaitForDocumentReadyStateToBeInteractiveOrBetter 'document.readyState == "complete"', timeout=timeout) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/browser/web_contents.py", line 239, in WaitForJavaScriptCondition return self._inspector_backend.WaitForJavaScriptCondition(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 52, in traced_function return func(*args, **kwargs) File "/b/swarming/w/ir/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend.py", line 291, in WaitForJavaScriptCondition e.message + '\n' + debug_message) TimeoutException: Timed out while waiting 59s for IsJavaScriptExpressionTrue. Console output:
,
Sep 23 2017
We have a similar issue with timeout for Wikipedia. What I think is happening is that the html is huge due to incredibly verbose tag attributes. chrishtr@ thinks that there's an optimize option we can turn on (may need to add a switch/flag) that will minimize the attributes and reduce html. This optimization is not on by default currently due to bugs, but they may be bugs that don't affect this page set. Another solution may be to just increase the expected timeout for these tests. I didn't see a way to do this for a particular story, but I haven't had a chance to look thoroughly yet, and we could always add a way, or, we could do it for the whole benchmark.
,
Sep 23 2017
Issue 764543 is the Wikipedia issue.
,
Sep 27 2017
,
Sep 27 2017
,
Sep 30 2017
On brief review snapping simple demo pages and looking for oddity, I see:
* perspective-origin and transform-origin on all elements
* id set on all elements even if not needed
* empty <style></style> tags
* explicit height present in style for all elements
* one-line style like 'border: 1px solid black' explodes into verbose
border-<direction>-{style,width}. Many such attributes explode.
* style minimization has a fixed max iteration of 5, and could have
bugs or low hanging opportunity for improvement
Possible solutions:
* look into why/if we need {perspective,transform}-origin
* only put id on elements when they've an associated style or need
* don't output <style> tag if there are no styles
* don't output an explicit height for elements unless needed
* run a re-collapse pass to turn verbose border-like styles into their
more condensed equivalents
* explore alternate max iteration or algorithm tweaks
Most of the explosion seems to be the last bullet, re: minimization
not really minimizing as much as it could. Need to build more
complex baseline demo pages to explore further.
The verbose attribute explosion looks like next biggest impact.
,
Oct 2 2017
We can extend the timeout, but that would increase the cycle time of the benchmark on bots. I think we should try to optimize the size/speed of static snapshot first before trying to extend timeout.
,
Oct 3 2017
The following revision refers to this bug: https://chromium.googlesource.com/catapult/+/ab715cde376e3c1d7b65afd91eb0479ad974be50 commit ab715cde376e3c1d7b65afd91eb0479ad974be50 Author: Nghia Nguyen <nednguyen@google.com> Date: Tue Oct 03 20:41:50 2017 Move snap-it to catapult's top third_party project This CL also update project's README to include contribution guideline & add OWNERS file. Bug: chromium:768010 Change-Id: I06de6ad8e6d469788e84af35ac1a61c7b31115bd Reviewed-on: https://chromium-review.googlesource.com/698807 Reviewed-by: Walter Korman <wkorman@chromium.org> Commit-Queue: Ned Nguyen <nednguyen@google.com> [add] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/README.md [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/tests/tests.html [modify] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/telemetry/telemetry/internal/snap_page_util.py [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/style.css [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/tests/tests.js [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/assets/camera.svg [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/assets/camera48.png [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/LICENSE [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/HTMLSerializer.js [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/popup.html [add] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/OWNERS [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/content_script.js [delete] https://crrev.com/c474937fb3f4a0af5dc2ddc2ed4491f7c7b1b543/telemetry/third_party/snap-it/README.md [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/assets/camera128.png [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/popup.js [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/manifest.json [delete] https://crrev.com/c474937fb3f4a0af5dc2ddc2ed4491f7c7b1b543/telemetry/third_party/snap-it/README.chromium [rename] https://crrev.com/ab715cde376e3c1d7b65afd91eb0479ad974be50/third_party/snap-it/assets/camera16.png
,
Oct 4 2017
The following revision refers to this bug: https://chromium.googlesource.com/catapult/+/8a0453f9d064d046ea74e47e2a76bdd96d20029d commit 8a0453f9d064d046ea74e47e2a76bdd96d20029d Author: Walter Korman <wkorman@chromium.org> Date: Wed Oct 04 18:14:56 2017 Don't emit empty style tags. We still emit an empty style tag with id='${this.pseudoElementTestingStyleId}' in HTMLSerializer, but we can look at that separately. Bug: chromium:768010 Change-Id: If5c078ae68973b4c466d002c7a46fdc9e63f63cc Reviewed-on: https://chromium-review.googlesource.com/696484 Reviewed-by: Ned Nguyen <nednguyen@google.com> Commit-Queue: Walter Korman <wkorman@chromium.org> [modify] https://crrev.com/8a0453f9d064d046ea74e47e2a76bdd96d20029d/third_party/snap-it/tests/tests.js [modify] https://crrev.com/8a0453f9d064d046ea74e47e2a76bdd96d20029d/third_party/snap-it/HTMLSerializer.js [modify] https://crrev.com/8a0453f9d064d046ea74e47e2a76bdd96d20029d/third_party/snap-it/popup.js
,
Nov 6 2017
Two changes in CQ seem to dramatically shrink the size of the previously huge pages: https://chromium-review.googlesource.com/c/chromium/src/+/755497 https://chromium-review.googlesource.com/c/chromium/src/+/755541 It's possible there's a bug in the external image support logic added where we used to embed images (which would make html huge) and now we treat them as external. However, cursory review does still show some inline embedded image:data. Perhaps this is an improvement in any case as it would allow us to re-enable the disabled pages.
,
Dec 7 2017
Aforementioned changes were reverted as the number of files caused benchmark cycle time to grow to an unacceptable level. nednguyen@ is looking at adding archive support to allow us to reland updated snapshots. This issue should therefore block on that.
,
Dec 8 2017
Transferring these to vmpstr@ as owner of the involved benchmark. Note that the r_and_r benchmark is in consideration to be replaced (for perf regression monitoring purposes at least) by an updated repaint benchmark. More on that at issue 770800.
,
Feb 5 2018
,
Jan 16
,
Jan 16
|
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by bugdroid1@chromium.org
, Sep 22 2017