page_cycler_v2_site_isolation.basic_oopif failure on chromium.perf |
||||||||
Issue descriptionpage_cycler_v2_site_isolation.basic_oopif failing on 13 builders Builders failed on: - Linux Perf (2): https://build.chromium.org/p/chromium.perf/builders/Linux%20Perf%20%282%29 - Mac 10.10 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Mac%2010.10%20Perf%20%282%29 - Mac 10.11 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Mac%2010.11%20Perf%20%282%29 - Mac HDD Perf (2): https://build.chromium.org/p/chromium.perf/builders/Mac%20HDD%20Perf%20%282%29 - Mac Retina Perf (2): https://build.chromium.org/p/chromium.perf/builders/Mac%20Retina%20Perf%20%282%29 - Win 10 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%2010%20Perf%20%282%29 - Win 7 ATI GPU Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%207%20ATI%20GPU%20Perf%20%282%29 - Win 7 Intel GPU Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%207%20Intel%20GPU%20Perf%20%282%29 - Win 7 Nvidia GPU Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%207%20Nvidia%20GPU%20Perf%20%282%29 - Win 7 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%207%20Perf%20%282%29 - Win 7 x64 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%207%20x64%20Perf%20%282%29 - Win 8 Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%208%20Perf%20%282%29 - Win 8 Zenbook Perf (2): https://build.chromium.org/p/chromium.perf/builders/Win%20Zenbook%20Perf%20%282%29
,
Aug 31 2016
Friendly sheriff ping
,
Aug 31 2016
Looks like a crash of some sort. Started happening consistently somewhere within 414704 and 414718 (https://chromium.googlesource.com/chromium/src/+log/61e97d3dd..08cd9d655c?pretty=full). Best guesses from the changelog: DevTools: merge devtools target with devtools host, part 1 https://codereview.chromium.org/2263843002 Add PlatformScreenDelegate and start implementation. https://codereview.chromium.org/2274353003 If there's no obvious culprit, I'll kick off a bisect and revert the offending CL.
,
Aug 31 2016
Actually the intersection of blame-range from all broken bots is 414705..414708 (https://chromium.googlesource.com/chromium/src/+log/16c88dfc4060..166a7cf0a05?pretty=full), so it's almost certainly one of the following 3 CLs: DevTools: merge devtools target with devtools host, part 1 https://codereview.chromium.org/2263843002 Bundle fake Widevine component manifest for stub CDM https://codereview.chromium.org/2136983002 Pavel, can you please take a look to see if this is indeed your CL causing the failures ASAP?
,
Aug 31 2016
(Sorry - one of these 2, the 3rd is iOS-only).
,
Aug 31 2016
Pavel says it was indeed probably his CL and he'll take a look later today or tomorrow. Given that it's only the site_isolation tests I think it's fine to wait another day or so. But there's a LOT of redness on the tree that's hard to wade through, so if it's not fixed in a day or two we should either revert the CL or disable the test.
,
Aug 31 2016
@Ned: Annie pointed me your way. Could you help me reproducing this? I'll fix it, but I need the repro.
,
Aug 31 2016
To reproduce locally, you can run: ./tools/perf/run_benchmark page_cycler_v2_site_isolation.basic_oopif --browser=release
,
Sep 1 2016
I can't get to the bottom of it. Could you tell me where telemetry fails and what expectations are not met?
,
Sep 1 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/7e4d11e55c607caad099671f2d46aa7726b65e99 commit 7e4d11e55c607caad099671f2d46aa7726b65e99 Author: pfeldman <pfeldman@chromium.org> Date: Thu Sep 01 02:53:26 2016 DevTools: make telemetry happy via calling oopifs 'iframes' in the discovery. BUG= 641528 TBR= Review-Url: https://codereview.chromium.org/2301633003 Cr-Commit-Position: refs/heads/master@{#415864} [modify] https://crrev.com/7e4d11e55c607caad099671f2d46aa7726b65e99/content/browser/devtools/devtools_agent_host_impl.cc
,
Sep 1 2016
Based on the log, telemetry fails at the step of closing the tab:
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/page/shared_page_state.py", line 228, in WillRunStory
self.browser.tabs[-1].Close()
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/browser/tab.py", line 100, in Close
self._tab_list_backend.CloseTab(self.id)
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome/tab_list_backend.py", line 66, in CloseTab
util.WaitFor(lambda: tab_id not in self.IterContextIds(), timeout=5)
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/core/util.py", line 86, in WaitFor
res = condition()
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome/tab_list_backend.py", line 66, in <lambda>
util.WaitFor(lambda: tab_id not in self.IterContextIds(), timeout=5)
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend_list.py", line 86, in IterContextIds
self._Update()
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/inspector_backend_list.py", line 99, in _Update
backends_map = self._devtools_client.GetUpdatedInspectableContexts()
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_client_backend.py", line 298, in GetUpdatedInspectableContexts
contexts = self._ListInspectableContexts()
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_client_backend.py", line 223, in _ListInspectableContexts
return self._devtools_http.RequestJson('')
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_http.py", line 107, in RequestJson
return json.loads(self.Request(path, timeout))
File "/b/c/b/Linux_Perf__2_/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_http.py", line 88, in Request
response = self._conn.getresponse()
File "/usr/lib/python2.7/httplib.py", line 1051, in getresponse
response.begin()
File "/usr/lib/python2.7/httplib.py", line 415, in begin
version, status, reason = self._read_status()
File "/usr/lib/python2.7/httplib.py", line 379, in _read_status
raise BadStatusLine(line)
To close the tab, telemetry sends a GET request like 127.0.0.1:<devtool_port>/json/close/<tab_id>. (<tab_id> is the id of context from devtool's response to 127.0.0.1:<devtool_port>/json/, this is devtools_http.RequestJson('') with |path| param equals to '', code is in https://cs.chromium.org/chromium/src/third_party/catapult/telemetry/telemetry/internal/backends/chrome_inspector/devtools_http.py?rcl=0&l=96)
After sending that close tab request, telemetry does a polling loop that keep calling devtools_http.RequestJson('') & check whether the return contexts dict from devtool no longer have the context id that it supposed to close. However, this fail as some point due to a bad response message from devtool:
File "/usr/lib/python2.7/httplib.py", line 1051, in getresponse
response.begin()
File "/usr/lib/python2.7/httplib.py", line 415, in begin
version, status, reason = self._read_status()
File "/usr/lib/python2.7/httplib.py", line 379, in _read_status
raise BadStatusLine(line)
Note that you can add more logging to browser to figure out what's going on when devtools response with that bad message. The way to do that is to rerun the benchmark command in #8 with "--browser-logging-verbosity=verbose".
,
Sep 1 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/71ff18e2385d68b077b081f2c98d8ee9c9b42108 commit 71ff18e2385d68b077b081f2c98d8ee9c9b42108 Author: pfeldman <pfeldman@chromium.org> Date: Thu Sep 01 18:43:21 2016 DevTools: check for target being oopif prior to matching its web contents. BUG= 641528 TBR=dgozman Review-Url: https://codereview.chromium.org/2303823002 Cr-Commit-Position: refs/heads/master@{#416006} [modify] https://crrev.com/71ff18e2385d68b077b081f2c98d8ee9c9b42108/chrome/browser/devtools/chrome_devtools_manager_delegate.cc
,
Sep 2 2016
The run is green again: https://build.chromium.org/p/chromium.perf/builders/Linux%20Perf%20%282%29
,
Sep 7 2016
Issue 643457 has been merged into this issue. |
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by martiniss@chromium.org
, Aug 26 2016