mesos-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ben Mahler" <benjamin.mah...@gmail.com>
Subject Re: Review Request: Initial support for CPU limits via CFS Bandwidth Control.
Date Sun, 03 Mar 2013 01:26:56 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/9464/#review17308
-----------------------------------------------------------


Getting close! A few high level comments:

1. I don't think we want to split the handlers, I thought it was cleaner when you were doing
all the cpu handling inside cpusChanged(). With this change there are now two handlers and
one uses the other. I think to keep things intuitive let's just place the logic inside cpusChanged
and guard it with a check of the flag.

2. What else do you anticipate using this flag for? I would consider it to be a temporary
measure until we can notify schedulers of the level of cpu isolation available. Hence wanting
it to be something simple like --cgroups_enable_cfs. We will then be killing this flag when
we make resources first class: https://issues.apache.org/jira/browse/MESOS-338.

- Ben Mahler


On March 2, 2013, 12:49 a.m., David Mackey wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/9464/
> -----------------------------------------------------------
> 
> (Updated March 2, 2013, 12:49 a.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Vinod Kone, and Ben Mahler.
> 
> 
> Description
> -------
> 
> Initial support for CPU "hard limits" via CFS Bandwidth Control in cgroups.
> 
> CFS is unique relative to existing Mesos cgroups support in that it is a "subfeature"
of an already supported cgroups subsystem, cpu. Also, there are two "tunables" for configuring
CFS bandwidth limiting.
> 
> There are 4 approaches one could take:
> 1) Use the CFS bandwidth limiting if the feature is present
> 2) Expose as separate flag, eg "cpu,cfs,memory,freezer"
> 3) Add feature flag support to subsystems via an additional delimiter, eg "cpu+cfs,memory,freezer".

> 4) Add an additional control flag via some other means
> 
> Option 2's downside breaks the 1:1 mapping between cgroups subsystems and a cgroups resource
flag.
> Option 3's downside is it greatly increases complexity of parsing cgroups subsystem flags.
> 
> This diff takes option 1. 
> 
> 
> This addresses bug MESOS-315.
>     https://issues.apache.org/jira/browse/MESOS-315
> 
> 
> Diffs
> -----
> 
>   src/slave/cgroups_isolation_module.hpp a04fc46 
>   src/slave/cgroups_isolation_module.cpp a779de8 
>   src/slave/flags.hpp d4aa045 
> 
> Diff: https://reviews.apache.org/r/9464/diff/
> 
> 
> Testing
> -------
> 
> make check + additional testing
> 
> 
> Thanks,
> 
> David Mackey
> 
>


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