> On Nov. 16, 2012, 7:09 p.m., Ben Mahler wrote:
> > src/slave/process_based_isolation_module.cpp, line 291
> > <https://reviews.apache.org/r/8077/diff/2/?file=190545#file190545line291>
> >
> >     Do you see a way to avoid introducing this boolean?
> 
> Vinod Kone wrote:
>     Hmmm, not sure how. The thing is an executor process terminated signal 
> might still need a proper clean up (for eg: process group in process based 
> isolation module or cgroup in cgroup based isolation module). Also, note that 
> killExecutor() can be called by the slave. Fwiw, the boolean 'killed' here is 
> used similarly in cgroups isolation module.

Ah, yes, I had a bit of deja-vu when looking at this ;)

To me it seems like killExecutor() should be smart enough to be called several 
times, so that we can avoid having to introduce more state (i.e. the killed 
boolean).
I could be convinced otherwise though, since I don't know this code very well.
I guess one advantage of ensuring it's only called once is that we can LOG any 
redundant/invalid kill attempts?


- Ben


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/8077/#review13523
-----------------------------------------------------------


On Nov. 15, 2012, 11:54 p.m., Vinod Kone wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/8077/
> -----------------------------------------------------------
> 
> (Updated Nov. 15, 2012, 11:54 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman and Ben Mahler.
> 
> 
> Description
> -------
> 
> This paves the way for a slave to shutdown executors that are not responsive 
> during launch (e.g. HDFS hang during fetching executor).
> 
> This fix is not enough for cgroups isolation module (enough for the process 
> based isolation module) because it currently blocks while launching an 
> executor.
> 
> 
> This addresses bug MESOS-310.
>     https://issues.apache.org/jira/browse/MESOS-310
> 
> 
> Diffs
> -----
> 
>   src/master/hierarchical_allocator_process.hpp 
> 5bcbeb99c823c3dbca0f89e19473b943c7469dab 
>   src/master/master.cpp 87fad0ed1eaecd59f2814d3e776b746b754b79b7 
>   src/slave/constants.hpp ddf02570caf3793106b3c48e158a5bb48c1ae80c 
>   src/slave/constants.cpp 1735a6b55a93e6537a5a119e5345961f3d84a000 
>   src/slave/process_based_isolation_module.hpp 
> efe59ebc0e8120926ea9f36b9eaa2f0b25830faf 
>   src/slave/process_based_isolation_module.cpp 
> 16fd584e78db2c517d828f2576ab8a38c5ce57ad 
>   src/slave/reaper.cpp c0ee4b4c07fd792bcb39455b666808b712eb32c2 
>   src/slave/slave.hpp bbba4404e9e2b1ff1e246f017cdad704438973ba 
>   src/slave/slave.cpp 7deb4574943aae4cfc5da5d6b3f600042686975f 
>   src/tests/master_tests.cpp 948ab5dff34eeba1f3ce593a864ddf282c8b19ed 
> 
> Diff: https://reviews.apache.org/r/8077/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Vinod Kone
> 
>

Reply via email to