New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 858963 link

Starred by 4 users

Issue metadata

Status: Fixed
Owner:
Closed: Jun 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 1
Type: Bug-Regression

Blocking:
issue 782002
issue 858922



Sign in to add a comment

chromedriver_py_tests failing 25% of the time on win7_chromium_rel_ng because tabs are crashing

Project Member Reported by kbr@chromium.org, Jun 29 2018

Issue description

I noticed a couple of tests fail in the chromedriver_py_tests suite in https://chromium-review.googlesource.com/1119622 and did a spot check on the bot.

https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng?limit=200 has failed in 50 of the last 200 CLs on the bot.

Just a few examples:

https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27609
https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27607
https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27779
https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27778
https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27602

I'm pretty sure all of these CLs are unrelated to the failing tests.

Looking at a recent failure:

https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/27781

these two tests failed:

__main__.ChromeDriverTest.testSlowIFrame
__main__.ChromeDriverTest.testUnexpectedAlertOpenExceptionMessage

here's the failure log from one of them:

ERROR: testSlowIFrame (__main__.ChromeDriverTest)
Verify ChromeDriver waits for slow frames to load.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "../../chrome/test/chromedriver/test/run_py_tests.py", line 370, in setUp
    self._driver = self.CreateDriver()
  File "../../chrome/test/chromedriver/test/run_py_tests.py", line 292, in CreateDriver
    **kwargs)
  File "e:\b\s\w\ir\chrome\test\chromedriver\test\..\client\chromedriver.py", line 237, in __init__
    response = self._ExecuteCommand(Command.NEW_SESSION, params)
  File "e:\b\s\w\ir\chrome\test\chromedriver\test\..\client\chromedriver.py", line 291, in _ExecuteCommand
    raise _ExceptionForLegacyResponse(response)
SessionNotCreatedException: session not created exception
from tab crashed
  (Session info: chrome=69.0.3477.0)
  (Driver info: chromedriver=2.40.571409 (9b2c83b74be21fec5fe66818e142de14d81ba02f),platform=Windows NT 6.1.7601 SP1 x86_64)


Basically – it looks like tab crashes are severely affecting this test suite. Unfortunately it doesn't look like minidumps are being generated and symbolized by this test harness.

This would be an ideal job for FindIt if it were handling this test suite.

Unfortunately the flakiness goes back to the beginning of the 200 builds on this bot so it looks like it's going to be difficult to track down whatever introduced this flakiness.

(Could this be related to the recent changes to these tests to handle retries?)

 

Comment 1 by st...@chromium.org, Jun 29 2018

Owner: johnchen@chromium.org
Status: Assigned (was: Untriaged)
Labels: -Sheriff-Chromium
Cc: cwinstanley@google.com johnchen@chromium.org
 Issue 859079  has been merged into this issue.
Status: Started (was: Assigned)
Status: Fixed (was: Started)
The culprit is r571282, which has been reverted with r571462. No more flakiness of chromedriver_py_tests are observed on win7_chromium_rel_ng after the revert.
Cc: crouleau@chromium.org
kbr@, is John's conclusion in comment 5 consistent with your understanding of the flakiness? I see that you wrote "the flakiness goes back to the beginning of the 200 builds on this bot".

Comment 8 by kbr@chromium.org, Jun 29 2018

Blocking: 782002
Cc: grunell@chromium.org sunn...@chromium.org
Yes, thanks. win7_chromium_rel_ng looks better at this point; no chromedriver_py_tests flakes.

Let's please keep these bugs linked together. Blocking the bug which added that CL. Also, a link to one of the failures of WebViewTest.SelectShowHide should have been added to either the revert CL or the original bug.

Comment 9 by kbr@chromium.org, Jun 29 2018

Blocking: 858922

Comment 10 by yigu@chromium.org, Jun 29 2018

 Issue 859188  has been merged into this issue.
Cc: kbr@chromium.org yihongg@chromium.org
 Issue 858922  has been merged into this issue.
Blocking: 864205
Blocking: -864205

Sign in to add a comment