mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Guangya Liu <gyliu...@gmail.com>
Subject Re: Review Request 50677: Renamed `OfferedResources` to `Allocation`.
Date Fri, 05 Aug 2016 21:51:24 GMT


> On 八月 2, 2016, 12:12 a.m., Benjamin Mahler wrote:
> > src/tests/hierarchical_allocator_tests.cpp, lines 3461-3466
> > <https://reviews.apache.org/r/50677/diff/1/?file=1459446#file1459446line3461>
> >
> >     Can we re-use the Allocation struct from above?
> >     
> >     https://github.com/apache/mesos/blob/9c6097f063405279efc07eec22457c2059653f07/src/tests/hierarchical_allocator_tests.cpp#L79-L83
> 
> Guangya Liu wrote:
>     Hi Ben, after some thought, seems here using `OfferedResources` maybe better, as
here we are actually constructing the `offers` on each agent; But the `Allocation`  is an
allocator concept and it means the resources for a specified framework.
>     
>     Also using `OfferedResources` can make the API call `recoverResources` more simple
as we can get the `slaveId` and `resources` on the agent directly from the `OfferedResources`;
If using `Allocation`, then we will need to traverse the hashmap to get the `slaveId` and
`resources` before `recoverResources`.
>     
>     The `benchmark` test is focusing on `offers`, so here we are using `OfferedResources`
in `benchmark` test; For other unit test cases other than `benchmark` test, they are focusing
on `allocations` for each framework, so it is using `Allocation`.
>     
>     What do you think?
> 
> Guangya Liu wrote:
>     Had some offline discussion with Ben, he also agreed to use `OfferedResources` for
all of the benchmark test, will update all of the benchmark test using the same concept to
make sure they are consistent.

Posted another patch https://reviews.apache.org/r/50868/ here to update `HierarchicalAllocator_BENCHMARK_Test.SuppressOffers`
to use `OfferedResources.`


- Guangya


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/50677/#review144412
-----------------------------------------------------------


On 八月 1, 2016, 10:59 p.m., Guangya Liu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/50677/
> -----------------------------------------------------------
> 
> (Updated 八月 1, 2016, 10:59 p.m.)
> 
> 
> Review request for mesos, Benjamin Mahler and Jiang Yan Xu.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This is a follow up action for https://reviews.apache.org/r/49616/ ,
> renaming `fferedResources` to `Allocation` would match the pattern
> used in the HierarchicalAllocatorTestBase.
> 
> 
> Diffs
> -----
> 
>   src/tests/hierarchical_allocator_tests.cpp e5576023e4aab930f0e2b1d82d8baa6569d852eb

> 
> Diff: https://reviews.apache.org/r/50677/diff/
> 
> 
> Testing
> -------
> 
> make
> make check
> 
> ```
> [ RUN      ] SlaveAndFrameworkCount/HierarchicalAllocator_BENCHMARK_Test.ResourceLabels/1
> Using 1000 agents and 50 frameworks
> Added 50 frameworks in 5231us
> Added 1000 agents in 3.295297secs
> round 0 allocate() took 2.323202secs to make 1000 offers after filtering 1000 offers
> round 1 allocate() took 2.326269secs to make 1000 offers after filtering 2000 offers
> round 2 allocate() took 2.393019secs to make 1000 offers after filtering 3000 offers
> round 3 allocate() took 2.33054secs to make 1000 offers after filtering 4000 offers
> round 4 allocate() took 2.388765secs to make 1000 offers after filtering 5000 offers
> round 5 allocate() took 2.364473secs to make 1000 offers after filtering 6000 offers
> round 6 allocate() took 2.631534secs to make 1000 offers after filtering 7000 offers
> round 7 allocate() took 2.692875secs to make 1000 offers after filtering 8000 offers
> round 8 allocate() took 2.653584secs to make 1000 offers after filtering 9000 offers
> ```
> 
> ```
> [ RUN      ] SlaveAndFrameworkCount/HierarchicalAllocator_BENCHMARK_Test.DeclineOffers/1
> Using 1000 agents and 50 frameworks
> Added 50 frameworks in 6797us
> Added 1000 agents in 2.628263secs
> round 0 allocate() took 2.207578secs to make 1000 offers after filtering  1000 offers
> round 1 allocate() took 2.230511secs to make 1000 offers after filtering  2000 offers
> round 2 allocate() took 2.238138secs to make 1000 offers after filtering  3000 offers
> round 3 allocate() took 2.229707secs to make 1000 offers after filtering  4000 offers
> round 4 allocate() took 2.239294secs to make 1000 offers after filtering  5000 offers
> round 5 allocate() took 2.310338secs to make 1000 offers after filtering  6000 offers
> round 6 allocate() took 2.30661secs to make 1000 offers after filtering  7000 offers
> round 7 allocate() took 2.377254secs to make 1000 offers after filtering  8000 offers
> round 8 allocate() took 2.399587secs to make 1000 offers after filtering  9000 offers
> round 9 allocate() took 2.540271secs to make 1000 offers after filtering  10000 offers
> ```
> 
> 
> Thanks,
> 
> Guangya Liu
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message