mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexander Rukletsov <ruklet...@gmail.com>
Subject Re: Review Request 43880: Added allocator metrics for total and allocated scalar resources.
Date Fri, 04 Mar 2016 10:02:01 GMT


> On March 3, 2016, 3:04 p.m., Alexander Rukletsov wrote:
> > src/master/allocator/mesos/hierarchical.cpp, lines 1711-1712
> > <https://reviews.apache.org/r/43880/diff/13/?file=1279033#file1279033line1711>
> >
> >     My intuition is that using `allocationScalarQuantities` will be more efficient.
However, we should expose known clients from the sorter. What do you think?
> 
> Benjamin Bannier wrote:
>     As for efficiency, I don't think there's an issue here; we just create a `Value::Scalar`
which I think cannot be avoided, and e.g., never create temporary `Resources`. Would we use
`allocationScalarQuantities`, we'd always be forced to create temporary `Resources` (which
in an ideal world could be optimized away again).

I was thinking that iterating over active roles should be cheaper than over all agents. But
it's fine to leave it as is and come back if we observe some performance issues.


- Alexander


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43880/#review121860
-----------------------------------------------------------


On March 3, 2016, 4:17 p.m., Benjamin Bannier wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/43880/
> -----------------------------------------------------------
> 
> (Updated March 3, 2016, 4:17 p.m.)
> 
> 
> Review request for mesos, Alexander Rukletsov and Ben Mahler.
> 
> 
> Bugs: MESOS-4720
>     https://issues.apache.org/jira/browse/MESOS-4720
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Added allocator metrics for total and allocated scalar resources.
> 
> 
> Diffs
> -----
> 
>   docs/monitoring.md 827f7073204fcf8575ca980a5571c8be4f5e4110 
>   src/master/allocator/mesos/hierarchical.hpp 3043888630b066505410d3b32c5b3f813cc458c1

>   src/master/allocator/mesos/hierarchical.cpp 70291075c00a9a557529c2562dedcfc6c6c3ec32

>   src/master/allocator/mesos/metrics.hpp PRE-CREATION 
>   src/master/allocator/mesos/metrics.cpp PRE-CREATION 
>   src/tests/hierarchical_allocator_tests.cpp 3e4ad31925e1b815a74d67fa3962d23fa5bc89d1

> 
> Diff: https://reviews.apache.org/r/43880/diff/
> 
> 
> Testing
> -------
> 
> make check (OS X)
> 
> I confirmed that this does not lead to general performance regressions in the allocator;
this is partially expected since the added code only inserts metrics in the allocator while
the actual work is perform asynchronously. These tests where performed with `HierarchicalAllocator_BENCHMARK_Test.DeclineOffers`
on an optimized build under OS X using clang(trunk) as compiler.
> 
> 
> Thanks,
> 
> Benjamin Bannier
> 
>


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