(No comment was entered for this change.)

Sorry, another one for you - can you take a look please? Thanks

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=56890 

------------------------------------------------------------------------
r56890 | rsesek@chromium.org | 2010-08-20 11:18:52 -0700 (Fri, 20 Aug 2010) | 12 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/startup/shutdown_test.cc?r1=56890&r2=56889
   M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/startup/startup_test.cc?r1=56890&r2=56889

Mark a bunch of startup_tests as FLAKY on Mac:
* ShutdownTest.SimpleSessionEnding
* ShutdownTest.TwentyTabsUserQuit
* StartupTest.PerfRestoreFewTabs
* StartupTest.PerfRestoreSeveralTabs
* StartupTest.PerfExtensionContentScript50

R=thomasvl
BUG=44067,52858
TEST=startup_tests

Review URL: http://codereview.chromium.org/3142024
------------------------------------------------------------------------

(No comment was entered for this change.)

This as a meta bug, so the priorities need to be set on the leaf bugs also

This is the master bug about figuring out the cause of failure in WaitForInitialLoads().  Paweł has a good idea for maybe increasing the robustness in  issue 52746 , but tracking down the root cause I think is more important.  The other sub-bugs are about specific instances of this kind of failure and I think are fine as P2, at least for now.

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=57089 

------------------------------------------------------------------------
r57089 | thomasvl@chromium.org | 2010-08-23 12:15:16 -0700 (Mon, 23 Aug 2010) | 5 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/startup/shutdown_test.cc?r1=57089&r2=57088

Mark another test as flaky on mac to try and keep the bot greener

BUG=52858
TEST=greener tree
Review URL: http://codereview.chromium.org/3201007
------------------------------------------------------------------------

I don't have the root cause on this yet, but playing with this some more...

The bundle check seems to always be returning the correct thing, and it's looking at the right executable.  But, things are actually running things slightly different then how they run when chrome is running:

base_paths_mac.cc uses AmIBundled() to decide how to walk from the executable to the source dir.  It uses the bundle flag to assume Chromium.app vs. unittest, so this isn't actually right for the Chromium Helper.app that's in the app bundle.

chrome_paths.cc uses AmIBundled() to find the log dir and the resources file. But these seem like that are probably right.

When browser_tests runs, the values in chrome_constants gets used to launch the main app with --type=renderer, that's why the icons show up in the docs instead of being hidden behind the scenes.  It really should use the helper to make sure everything is the same.  This probably extends to any test that directly triggers workers, etc.

Do any of these cause the problems - not likely, I'd expect this to be a total failure, not flaky.

Mark - you did the framework/helper, do you think it's worth trying to correct this type of plumbing?  the base_paths_mac part will be ugly since bundle alone won't be enough to say for sure.

I updated bug 46609 the other day, WaitForInitialLoads() failures do happen on windows (no where near as often), but I'm wondering if there is something about timings and/or runloops/ipc that makes this just easier to get on the Mac.

We have a bug somewhere about asserting if the main app gets launched with --type, or the helper gets launched with an unknown --type.

It’s even more important to only test known .app bundle/--type combinations.

A test that launches the main app bundle but uses --type isn’t testing exactly what we think it is.

 bug 23645  is for detecting the use of type on the main bundle.  i'll see about a cl to enforce that.

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=57231 

------------------------------------------------------------------------
r57231 | thomasvl@chromium.org | 2010-08-24 13:50:06 -0700 (Tue, 24 Aug 2010) | 8 lines
Changed paths:
   M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/app/chrome_dll_main.cc?r1=57231&r2=57230
   M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/in_process_browser_test.cc?r1=57231&r2=57230

[Mac] Enforce proper usage of main app bundle and helper app bundle.
- Add a check to make sure the helper isn't used without --type
- Add a check to make sure the main app isn't used with --type.
- Make inproc browser tests invoke the helper that is within the app bundle on disk to recreate the real world environment.

BUG=23645,52858
TEST=green tree, invoking the main app with --type fails, and invoking the helper without it fails.
Review URL: http://codereview.chromium.org/3171027
------------------------------------------------------------------------

A very large number of tests are now flaky because of this problem, to the point where I haven't been bothering to mark them as such.  I've just been checking each failure manually, because that seemed less silly than marking 50% of our tests as flaky :)

Quite a lot of tests fail with similar warnings on the TSan/FYI bot since
http://build.chromium.org/buildbot/waterfall.fyi/builders/Linux%20Tests%20(tsan%20ui)/builds/1827

I think this may be related.

I am convinced now that this is the same root problem as in  Issue 35374 .  The renderer is getting wedged (deadlocked?) very soon after the fork, so it is unable to exec() and actually accomplish anything.  Without a renderer, we are unable to perform the initial load.

I confirmed this by adding a bunch of logging in r58576 to see if we were never returning from the MachPortSender constructor.  The actual offending call is bootstrap_look_up(), which apparently never returns.

Example logs are pasted below.  Notice how we try to create the sender port but never print the "done creating" message.

PluginTest.Flash: 
[47467:22275:3478590562828178:ERROR:/b/slave/chromium-rel-mac-builder/build/src/base/process_util_posix.cc(330)] fork_and_get_task() about to fork
[47467:22275:3478590563920769:ERROR:/b/slave/chromium-rel-mac-builder/build/src/base/process_util_posix.cc(358)] fork_and_get_task() I'm the parent!
[47469:515:3478590566268893:ERROR:/b/slave/chromium-rel-mac-builder/build/src/base/process_util_posix.cc(336)] fork_and_get_task() I'm the child!
[47469:515:3478590566458401:ERROR:/b/slave/chromium-rel-mac-builder/build/src/base/process_util_posix.cc(346)] fork_and_get_task() child creating sender port
[47467:22275:3478590664212000:ERROR:/b/slave/chromium-rel-mac-builder/build/src/base/process_util_posix.cc(362)] parent WaitForMessage() failed: 0x10004003 (ipc/rcv) timed out
[47466:2307:0903/204835:3478615328759031:ERROR:/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc(299)] WaitForInitialLoadsFailed
[47466:2307:0903/204835:3478615330418466:ERROR:/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc(278)] Launching browser again, retry #1
[47470:22787:3478615416139223:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(885)] pipe error on 3 Currently writing message of size:24: Socket operation on non-socket
[47470:22787:3478615416198435:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(885)] pipe error on 3 Currently writing message of size:24: Socket operation on non-socket
[47470:22787:3478615416246918:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(500)] pipe error (3): Socket operation on non-socket
[47466:2307:0903/204900:3478640335964039:ERROR:/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc(299)] WaitForInitialLoadsFailed
[47466:2307:0903/204900:3478640337599630:ERROR:/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc(278)] Launching browser again, retry #2
[47472:22787:3478640427478630:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(885)] pipe error on 3 Currently writing message of size:24: Socket operation on non-socket
[47472:22787:3478640427525865:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(885)] pipe error on 3 Currently writing message of size:24: Socket operation on non-socket
[47472:22787:3478640427580856:ERROR:/b/slave/chromium-rel-mac-builder/build/src/ipc/ipc_channel_posix.cc(500)] pipe error (3): Socket operation on non-socket
[47466:2307:0903/204925:3478665340603092:ERROR:/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc(299)] WaitForInitialLoadsFailed
/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc:311: Failure
Failed
Failed to launch browser

(No comment was entered for this change.)

Given the fix in  bug 35374  i'm gonna close this.

Verified label updated by AutoAllocator, contact AmolK or KrisR for details

(No comment was entered for this change.)

(No comment was entered for this change.)

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=64264

------------------------------------------------------------------------
r64264 | stuartmorgan@chromium.org | Thu Oct 28 10:14:49 PDT 2010

Changed paths:
 M http://src.chromium.org/viewvc/chrome/trunk/src/base/debug/stack_trace_unittest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/browser/debugger/devtools_remote_listen_socket_unittest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome_frame/test/test_with_web_server.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/common/logging_chrome_uitest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/startup/shutdown_test.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/browser/metrics/metrics_service_uitest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/browser/renderer_host/resource_dispatcher_host_uitest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/ui/npapi_uitest.cc?r1=64264&r2=64263&pathrev=64264
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/startup/startup_test.cc?r1=64264&r2=64263&pathrev=64264

Update a bunch of FLAKY_ tests to match current reality

Mostly this is removing FLAKY_ where the bug was fixed but the label never removed, but also switches some FLAKY_ to platform-specific FAILS_.

BUG=28372,32048,32070,35341,45561,48544,48562,52858
TEST=N/A

Review URL: http://codereview.chromium.org/4087009
------------------------------------------------------------------------

The last 50 builds from r[56668, 56855] have had various tests fail on Mac10.6 Perf (2).  This is the failure message; it seems to be causing many of our UI tests to fail, not just perf ones:

/b/slave/chromium-rel-mac-builder/build/src/chrome/test/ui/ui_test.cc:278: Failure
Value of: server_->WaitForInitialLoads()
  Actual: false
Expected: true

The following is a list from Mac10.6 Perf(2) of the failing tests and the number of times each failed:
(11) StartupTest.PerfRestoreSeveralTabs
(7) ShutdownTest.TwentyTabsUserQuit
(6) ShutdownTest.SimpleSessionEnding
(6) StartupTest.PerfExtensionContentScript50
(6) StartupTest.PerfRestoreFewTabs
(5) ShutdownTest.SimpleWindowClose
(5) ShutdownTest.TwentyTabsSessionEnding
(4) ShutdownTest.TwentyTabsWindowClose
(4) StartupTest.PerfCold
(4) StartupTest.PerfFewTabs
(3) StartupTest.PerfComplexTheme
(2) ShutdownTest.SimpleUserQuit
(1) StartupTest.PerfExtensionContentScript1
(1) StartupTest.PerfExtensionEmpty

There are a bunch of other issues related to this kind of failure.  This will be the meta bug for WaitForInitialLoads().

The following revision refers to this bug:
    http://src.chromium.org/viewvc/chrome?view=rev&revision=121942

------------------------------------------------------------------------
r121942 | evan@chromium.org | Tue Feb 14 13:35:25 PST 2012

Changed paths:
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/perf/startup_test.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/ui/npapi_uitest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/ui/layout_plugin_uitest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/automation/dom_automation_browsertest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/ui/ppapi_uitest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/perf/feature_startup_test.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/reliability/automated_ui_test_test.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/automation/automation_proxy_uitest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/automation/extension_proxy_uitest.cc?r1=121942&r2=121941&pathrev=121942
 M http://src.chromium.org/viewvc/chrome/trunk/src/chrome/test/base/chrome_process_util_uitest.cc?r1=121942&r2=121941&pathrev=121942

Flakiness cleanup: disable flaky tests under chrome/test/

See https://groups.google.com/a/chromium.org/group/chromium-dev/browse_thread/thread/fcec09fc659f39a6

BUG=98071,109642,32293,60409,44617,62143,25039,72745,59441,59440,79175,69940,101591,46609,52858,102584,99604,96049,112821,96039,113406,81050,59327,30702,71904,44960,104380,111508,109258,111636,111355,111209,63239


Review URL: http://codereview.chromium.org/9358065
------------------------------------------------------------------------

This issue has been closed for some time. No one will pay attention to new comments.
If you are seeing this bug or have new data, please click New Issue to start a new bug.

(No comment was entered for this change.)