mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benno Evers <bev...@mesosphere.com>
Subject Re: Review Request 70884: Added optional 'host' string member to UPID.
Date Fri, 05 Jul 2019 13:44:56 GMT


> On July 3, 2019, 4:07 p.m., Benjamin Mahler wrote:
> > Was the benchmark done with an optimized build?
> > Can you expand on the performance implications of this change in the description?
(e.g. compare the min, 25th percentile, median, 75th percentile, max of at least 10 runs).

Sorry, I did not see this comment before committing. The benchmark above was done on my laptop,
I think using an optimized build but I can't prove it anymore. I can repeat the benchmark
on one of our build machines and record the results in some JIRA instead.


> On July 3, 2019, 4:07 p.m., Benjamin Mahler wrote:
> > 3rdparty/libprocess/include/process/pid.hpp
> > Lines 204 (patched)
> > <https://reviews.apache.org/r/70884/diff/4/?file=2153428#file2153428line204>
> >
> >     Hm.. hostname or host?

I opted for `host` here because that's the name that was already used in the PID-parsing code.
However, since this is an internal field there should be no problem updating the name if we
decide something else would be more appropriate.


- Benno


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/70884/#review216344
-----------------------------------------------------------


On June 26, 2019, 1:37 p.m., Benno Evers wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/70884/
> -----------------------------------------------------------
> 
> (Updated June 26, 2019, 1:37 p.m.)
> 
> 
> Review request for mesos, Benjamin Mahler, Joseph Wu, and Till Toenshoff.
> 
> 
> Bugs: MESOS-9858
>     https://issues.apache.org/jira/browse/MESOS-9858
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This allows client code to access the original hostname
> that was used to specify a libprocess address.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/include/process/pid.hpp 9f09ab46fa3ceaeac09b0fbf9f532728c4ed2d7a

>   3rdparty/libprocess/src/pid.cpp fdc61b5ab6c75b33ce33de7edd11e9302550f300 
> 
> 
> Diff: https://reviews.apache.org/r/70884/diff/4/
> 
> 
> Testing
> -------
> 
> Results from running the `Process_BENCHMARK_ThroughputPerformance` with `--gtest_repeat=10`.
There seems to be a slight degradation of performance on average, but the benchmark runtime
seems to be dominated by other factors, as indicated by the large variance and presence of
outliers on both sides.
> 
> 
> # Master
> 2519083.849973
> 2401163.451671
> 2322734.358626
> 2285763.292828
> 1883268.801707
> 2179436.004374
> 2071825.301714
> 2105588.897964
> 2067228.706673
> 
> 
> # Including UPID changes
> 2305142.571762
> 2194902.761484
> 2139113.533129
> 2024036.049603
> 1765327.993699
> 2172592.534808
> 2053217.096855
> 2103817.232509
> 1480080.596818
> 1808977.015807
> 
> 
> Thanks,
> 
> Benno Evers
> 
>


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