> On Jan. 18, 2016, 8:26 a.m., Till Toenshoff wrote:
> > src/tests/fetcher_cache_tests.cpp, line 842
> > <https://reviews.apache.org/r/42149/diff/4/?file=1200150#file1200150line842>
> >
> >     Given that pause() and resume() will always be used on the same thread, 
> > we dont have a race here accross an already used latch just after this 
> > case, right?

Right.


- Bernd


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/42149/#review115013
-----------------------------------------------------------


On Jan. 18, 2016, 3:28 a.m., Bernd Mathiske wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/42149/
> -----------------------------------------------------------
> 
> (Updated Jan. 18, 2016, 3:28 a.m.)
> 
> 
> Review request for mesos, Adam B, Alexander Rojas, Benjamin Hindman, Joseph 
> Wu, Till Toenshoff, and Timothy Chen.
> 
> 
> Bugs: MESOS-3235
>     https://issues.apache.org/jira/browse/MESOS-3235
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Also inlined the function that awaits fetch contention.
> 
> This mutex was prone to causing races at task startup by firmly
> blocking an internal libprocess thread. The latch avoids this.
> 
> Failing to launch a task due to such a race did not get flagged
> by directly related test failures, because the AWAIT catching this
> situation was ineffective, having been placed inside a call from the
> test. Only the subsequent wait for task completion triggered a test
> failure then. By then it was obscured what exactly had happened.
> 
> 
> Diffs
> -----
> 
>   src/tests/fetcher_cache_tests.cpp 1fb1e213d3c35479789688d1a3a49a3c6058b198 
> 
> Diff: https://reviews.apache.org/r/42149/diff/
> 
> 
> Testing
> -------
> 
> make check
> bin/mesos-tests.sh --gtest_repeat=1000 --gtest_break_on_failure 
> --gtest_filter="*HttpCachedConcurrent*"
> 
> 
> Thanks,
> 
> Bernd Mathiske
> 
>

Reply via email to