mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jojy Varghese" <j...@mesosphere.io>
Subject Re: Review Request 36106: cgroups: added cpuacct subsystem
Date Fri, 03 Jul 2015 00:31:33 GMT


> On July 2, 2015, 11:32 p.m., Ben Mahler wrote:
> > src/linux/cgroups.hpp, lines 443-472
> > <https://reviews.apache.org/r/36106/diff/1/?file=997646#file997646line443>
> >
> >     Thanks!
> >     
> >     (1) Do you mind updating my TODO on cgroups::stat() to reflect that cpuacct::stat
is implemented?
> >     
> >     (2) Can we just make this a simple struct with two non-const fields and remove
the parse method and getters? Keep it simple and small, if we want immutability we can just
have a 'const Stat' rather than forcing it on the caller :)
> >     
> >     (3) Any reason not to use 'Duration' for these fields?

1) Absolutely I can. 
2) I wanted to reflect the semantics of the stats call. When you make a stats call, the data
you get is immutable. By forcing external const, it would imply that the value is immutable.
3) Duration is a "period" ( i think) and the values here are absolute values derived from
ticks.


> On July 2, 2015, 11:32 p.m., Ben Mahler wrote:
> > src/linux/cgroups.hpp, lines 489-492
> > <https://reviews.apache.org/r/36106/diff/1/?file=997646#file997646line489>
> >
> >     Whoops, this duplicates cpu::cfs_quota_us() above!

whoops!


> On July 2, 2015, 11:32 p.m., Ben Mahler wrote:
> > src/linux/cgroups.cpp, lines 2014-2027
> > <https://reviews.apache.org/r/36106/diff/1/?file=997647#file997647line2014>
> >
> >     Any reason you're not just re-using cgroups::stat here? I'd suggest just calling
cgroups::stat for now, should simplify this and make it easier for the reader. :)

The only reason being that the way cpuacct creates Stat should be encapsulated in the cpuacct::Stat.
This is the same reason there is a parse method in Stat. But I can change it to use cgroups::stat
if absolutely necessary.


- Jojy


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/36106/#review90288
-----------------------------------------------------------


On July 2, 2015, 8:09 p.m., Jojy Varghese wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/36106/
> -----------------------------------------------------------
> 
> (Updated July 2, 2015, 8:09 p.m.)
> 
> 
> Review request for mesos, Ian Downes, Jie Yu, Joris Van Remoortere, and Timothy Chen.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> cgroups implementation does not have a cpuacct subsystem implementation as of
> today. Adding the implementation for stat function.
> 
> Changes:
>   - added Stats class to encapsulate cpuacct.stat data
>   - added implementation for cpuacct::stats
>   - added unit tests
> 
> Jira: MESOS-2961
> 
> 
> Diffs
> -----
> 
>   src/linux/cgroups.hpp 73b98317880eea3d6a2ba37ac56d1f7e3600ba94 
>   src/linux/cgroups.cpp 4c006d0c7382b940a83359d636c0d48952cdbb00 
>   src/tests/cgroups_tests.cpp 475f48a474eea708f98d8c0300862351a2d4379a 
> 
> Diff: https://reviews.apache.org/r/36106/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Jojy Varghese
> 
>


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