[chromium-dev] Fwd: Flakiness in net_unittests on Mac Debug

2008-12-23 Thread Jeremy Moskovich
Whoops, forgot to cc chromium-dev, also see the reply from dkegel...

-- Forwarded message --

So the CHECK() failing means that MessagePumpLibevent::Init() is returning
false.

This could be the result of failures in:
* Creating the pipe() (line 84)
* Setting either end to Non-blocking. (lines 86,88)
* event_add() failing (line 98).

Dan: any ideas?

This doesn't appear to be related [at least directly] to the refactoring I
did...

Best regards,
Jeremy

--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com 
View archives, change email options, or unsubscribe: 
http://groups.google.com/group/chromium-dev
-~--~~~~--~~--~--~---



[chromium-dev] Fwd: Flakiness in net_unittests on Mac Debug

2008-12-23 Thread Jeremy Moskovich
forwarding to chromium-dev...

-- Forwarded message --
From: Dan Kegel

That seems like a good guess to me.

I'm running net_unittests on the mac in a loop now
with log messages around each of the failure points.
It's been running for ten minutes without failure.

Maybe we should just check in the attached patch so we can learn more
next time it happens.

On Tue, Dec 23, 2008 at 10:40 AM, Jeremy Moskovich wrote:
 Too many open FDs causing pipe() to fail?

 Are any of you able to repro this locally?

 Best regards,
 Jeremy

 On Tue, Dec 23, 2008 at 10:38 AM, e. roman wrote:

 I don't know what this is either.

  the only change that seems to be related at all is r7229
  (http://codereview.chromium.org/14850). Pulling in Eric for that change

 r7229 was a no-op change (would have only resulted in a changed binary
 when compiling with the PURIFY flag on Windows), so I can't imagine
 that being responsible.

 Sounds like flakiness from another source.

 
  On Tue, Dec 23, 2008 at 11:46 AM, Avi Drissman  wrote:
 
  The net_unittests started being flaky due to hitting a check in
  message_pump_libevent:
 
  [ RUN  ] URLRequestTest.Post307RedirectPost
  /b/pylibs/zope/__init__.py:19: UserWarning: Module zope was already
  imported from /b/pylibs/zope/__init__.pyc, but
 
 
/System/Library/Frameworks/Python.framework/Versions/2.5/Extras/lib/python
  is being added to sys.path
 
 
import pkg_resources
 
 
 
[1222/150922:FATAL:/b/slave/sub-dbg-mac/build/src/base/message_pump_libevent.cc(79)]
  Check failed: false.
 
 
 
[1222/150922:FATAL:/b/slave/sub-dbg-mac/build/src/base/message_pump_libevent.cc(79)]
  Check failed: false.
 
  This has been sporadically failing for a while now (see
 
 
http://build.chromium.org/buildbot/waterfall/waterfall?builder=Modules%20Mac%20(dbg)http://build.chromium.org/buildbot/waterfall/waterfall?builder=Modules%20Mac%20%28dbg%29
  ). I'm still scrolling back to find the first time this happened, but
  you
  were the last one to make real changes here. Thoughts?
 
  Avi
 
 



--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com 
View archives, change email options, or unsubscribe: 
http://groups.google.com/group/chromium-dev
-~--~~~~--~~--~--~---



pump.patch
Description: Binary data