Issue metadata
Sign in to add a comment
|
5 pages of page_cycler_v2.typical_25 would fail using local build ChromiumOS image
Reported by
lingyun....@intel.com,
Jul 5
|
||||||||||||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36 Steps to reproduce the problem: (1)sync Chromium OS source code with manifest 10798.0.0-rc2.xml from paladin folder in manifest-versions repo (https://chromium.googlesource.com/chromiumos/manifest-versions) (2)sync Chromium to 69.0.3464.0 (3)build local image with below commands: 1) cros_sdk --enter --chrome_root=chrome_root 2) export CHROME_ORIGIN=LOCAL_SOURCE 3) export BOARD=reef 4) ./setup_board --board=$BOARD 5) cros_workon --board=$BOARD start chromeos-chrome 6) USE="afdo_use thinlto" ./build_packages --board=$BOARD 7) ./build_image --board=$BOARD --noenable_rootfs_verification test (4) run page_cycler_v2.typical_25 using local build ChromiumOS image and check the results What is the expected behavior? All pages are loaded successfully with metric values using local build ChromiumOS images What went wrong? 5 pages would fail: (1) economist (2) allrecipes (3) money.cnn (4) premierleague (5) ign However, no failure observed if using partner ChromeOS image with the same version (i.e., 10798.0.0). The error messages from local build ChromiumOS image are shown as: Traceback (most recent call last): File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/internal/story_runner.py", line 123, in _RunStoryAndProcessErrorIfNeeded state.RunStory(results) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 75, in traced_function return func(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/page/shared_page_state.py", line 311, in RunStory self._current_page.Run(self) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/page/__init__.py", line 100, in Run self.RunPageInteractions(action_runner) File "/home/test/chromium_telemetry_new_5/src/tools/perf/page_sets/typical_25.py", line 27, in RunPageInteractions 'performance.timing.loadEventStart > 0') File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 75, in traced_function return func(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/internal/actions/action_runner.py", line 261, in WaitForJavaScriptCondition return self._tab.WaitForJavaScriptCondition(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 75, in traced_function return func(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/internal/browser/web_contents.py", line 239, in WaitForJavaScriptCondition return self._inspector_backend.WaitForJavaScriptCondition(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/common/py_trace_event/py_trace_event/trace_event_impl/decorators.py", line 75, in traced_function return func(*args, **kwargs) File "/home/test/chromium_telemetry_new_5/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend.py", line 302, in WaitForJavaScriptCondition e.message + '\n' + debug_message) TimeoutException: Timed out while waiting 60s for IsJavaScriptExpressionTrue. Did this work before? Yes Chrome version: 69.0.3464.0 Channel: n/a OS Version: 10798.0.0 Flash Version:
,
Jul 17
,
Jul 17
Hi Kouhei, according to "Chrome Benchmark Reset", the owner of page_cycler series is you and kasakamoto. Would you mind to take a look?
,
Jul 20
Hi kouhei@ and ksakamoto@, would you mind to take a look and share your comments? Please let me know if any data needed, I could help to provide.
,
Jul 20
Sorry, I don't have much insight now, since we retired page_cycler_v2 but for ChromeOS.
,
Jul 20
,
Jul 20
+Chrome OS Perf team lead. Perhaps we should move to loading.desktop.
,
Jul 24
,
Jul 24
Please provide the full command of (4) run page_cycler_v2.typical_25, and attach the full test log. Thanks.
,
Jul 24
Full command of (4) is : ./run_benchmark --browser=cros-chrome --remote=$DUT_IP page_cycler_v2.typical_25 --pageset-repeat=1 --reset-results -v Test log from telemetry is also attached. I just ran one page 'ign.com' which could reproduce the issue.
,
Jul 30
Hi vovoy@, I was wondering if you have any update on this issue? Please let me know if any more data are needed.
,
Jul 30
,
Jul 30
page_cycler_v2 is deprecated, I will migrate it to loading.desktop on Chrome OS.
,
Jul 31
Hi lingyun-cai@, you can use the following benchmark for page loading time measuring. ./run_benchmark --browser=cros-chrome --remote=$DUT_IP --story-tag-filter=typical loading.desktop
,
Aug 1
Hi vovoy@, thanks for the update. We've tried loading.desktop on Chrome OS and it worked. We noticed that there are multiple metrics such as FP/FCP/FMP and timeToOnload, which one do you pay more attention to on regular perf tracking?
,
Aug 1
timeToFirstContentfulPaint: Time until something nontrivial (e.g. text or image) is painted for the first time. timeToFirstMeaningfulPaint: Time until the first paint where the page's primary content is visible (heuristically determined). FirstPaint: Time until anything is painted for the current document. timeToOnload: Time until the onload event. Because this has low correlation with user-perceived load time, we do not recommend using this as a key metric. I would pay more attention on timeToFirstContentfulPaint.
,
Aug 22
We recently found the possible cause of the failures on page_cycler_v2 using local ChromiumOS image, which is related to site isolation. On local build ChromiumOS, site isolation is enabled by default, while disabled by default on official ChromeOS. If we ran page_cycler_v2.typical_25 with Telemetry flag '--extra-browser-args="--disable-site-isolation-trials"' on local ChromiumOS, the failure pages would be loaded successfully.
,
Jan 11
This issue has an owner, a component and a priority, but is still listed as untriaged or unconfirmed. By definition, this bug is triaged. Changing status to "assigned". Please reach out to me if you disagree with how I've done this.
,
Jan 16
(6 days ago)
,
Jan 16
(6 days ago)
|
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by hong.zh...@intel.com
, Jul 5