Re: Review Request 43358: Added support for all additional Perf Events in PerfEventIsolator.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/43358/#review118637 --- src/linux/perf.cpp (line 474) <https://reviews.apache.org/r/43358/#comment179929> Shouldn't we use 'nullptr' instead of NULL? - Szymon Konefal On Feb. 10, 2016, 12:39 p.m., Bartek Plotka wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/43358/ > --- > > (Updated Feb. 10, 2016, 12:39 p.m.) > > > Review request for mesos, Ben Mahler, Chi Zhang, Ian Downes, Niklas Nielsen, > Paul Brett, Szymon Konefal, and Cong Wang. > > > Bugs: MESOS-4595 > https://issues.apache.org/jira/browse/MESOS-4595 > > > Repository: mesos > > > Description > --- > > Added dynamic map `extra_events` for all additional perf events which are not > specified explicitly `PerfStatistics` > TODO in next PRs: Add test and extend description for `--perf_events=` option > in Mesos flags. > > > Diffs > - > > include/mesos/mesos.proto 194750e92020753e60154083a47bdc3398d31466 > src/linux/perf.cpp 1c113a2b3f57877e132bbd65e01fb2f045132128 > > Diff: https://reviews.apache.org/r/43358/diff/ > > > Testing > --- > > make check > > > Thanks, > > Bartek Plotka > >
Re: Review Request 34662: Modularized ResourceEstimator and added test for that module
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/34662/#review85146 --- src/tests/oversubscription_tests.cpp <https://reviews.apache.org/r/34662/#comment136634> Why CHECK_SOME rather than ASSERT_SOME? CHECK_SOME prints only a log, and ASSERT_SOME could return ::testing::AssertionFailure() - Szymon Konefal On May 26, 2015, 3:55 a.m., Bartek Plotka wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/34662/ > --- > > (Updated May 26, 2015, 3:55 a.m.) > > > Review request for mesos, Jie Yu, Niklas Nielsen, Szymon Konefal, and Vinod > Kone. > > > Bugs: MESOS-2650 > https://issues.apache.org/jira/browse/MESOS-2650 > > > Repository: mesos > > > Description > --- > > Added *ResourceEstimator* (RE) module interface. > Added *TestResourceEstimator* example module. (Noop *ResourceEstimator* > module) > Moved *TestResourceEstimator* to seperate file from *tests/mesos.hpp* > Added *DummyTestResourceEstimator* module for unit tests purpose. (Module > with *TestResourceEstimator* logic - that's why it has to be in seperate file) > Changed *oversubscription_tests* to be typed_tested (for normal RE and RE > module) > > NOTE: The example modules were good enough for other modules' unit tests, > however RE had to be extended - to push particular *Resources* to slave. > That's why it was necessary to add new "Dummy" *TestResourceEstimator* module. > > In future if we implement https://issues.apache.org/jira/browse/MESOS-2764, > we will be able to inject stubed *Resources* in better way. > > > Diffs > - > > include/mesos/module/resource_estimator.hpp PRE-CREATION > src/Makefile.am 814468e3c5c750a6649b5eeb7c7f945f9e025c19 > src/examples/test_resource_estimator_module.cpp PRE-CREATION > src/module/manager.cpp 5fed1aa82f72aa55c8e4e946f5f458ee357162be > src/tests/mesos.hpp b8f7a2f9236166e42421d926718af8d45e857eba > src/tests/module.hpp c379f01ae4e185960b8710b6a10626a3b8784fc8 > src/tests/module.cpp c4f811998aef14ab247380318fab79f18672ab29 > src/tests/oversubscription_tests.cpp > 75c25b04c1e6a8e0e7e8fd55440743fe1699af88 > src/tests/resource_estimator.hpp PRE-CREATION > src/tests/resource_estimator_module.cpp PRE-CREATION > > Diff: https://reviews.apache.org/r/34662/diff/ > > > Testing > --- > > make check > > > Thanks, > > Bartek Plotka > >
Re: Review Request 34571: Pushed QoS Correction stub message to mesos.proto
> On May 21, 2015, 11:08 p.m., Niklas Nielsen wrote:
> > A higher level question is whether we should have touples of actions.
> >
> > For example:
> >
> > message KillAction {
> > optional ExecutorID executor_id = X;
> > }
> >
> > message ResizeAction {
> > optional TaskID task_id = X;
> > optional Resources resources = Y;
> > }
> >
> > ...
Yes, the resizing action of custom isolators could be tricky to formalize.
We have two choices:
1) Make an union of possible actions with bodies
2) Use labels to carry metadata
I would go for the first choice. If someone would like to resize his custom
iso, one should need to extend the ResiseAction message.
- Szymon
---
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/34571/#review84814
---
On May 21, 2015, 11:02 p.m., Bartek Plotka wrote:
>
> ---
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/34571/
> ---
>
> (Updated May 21, 2015, 11:02 p.m.)
>
>
> Review request for mesos, Jie Yu, Niklas Nielsen, and Vinod Kone.
>
>
> Bugs: MESOS-2760
> https://issues.apache.org/jira/browse/MESOS-2760
>
>
> Repository: mesos
>
>
> Description
> ---
>
> This part of proto describes a QoS corrections message which includes
> corrections for particular executors or tasks.
> It is a generic message between QoS Controller and slave.
>
>
> Diffs
> -
>
> include/mesos/mesos.proto a66888916dc90e476c8bb20e67e3f6b08c47fb99
>
> Diff: https://reviews.apache.org/r/34571/diff/
>
>
> Testing
> ---
>
> make check
>
>
> Thanks,
>
> Bartek Plotka
>
>
