mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Mann <g...@mesosphere.io>
Subject Re: Review Request 71886: Set container's `cpu.cfs_quota_us` to its CPU resource limit.
Date Fri, 28 Feb 2020 18:20:45 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/71886/#review219699
-----------------------------------------------------------




src/slave/containerizer/mesos/isolators/cgroups/subsystems/cpu.cpp
Lines 107 (patched)
<https://reviews.apache.org/r/71886/#comment307915>

    For executor containers, you have logic in `launchExecutor()` which inspects the value
of `flags.cgroups_enable_cfs`, and also sets the limits equal to the resource requests when
appropriate.
    
    It seems strange to me to have some of that logic in the agent code, and some of that
logic in the isolators. I would expect that we either put it all in the agent, or all in the
isolators.
    
    What about adding code to the `LaunchContainer` API handler which does this for tasks,
and then the isolator doesn't have to care about the semantics of the `cgroups_enable_cfs`
flag and the limits being set to the resource requests in some cases.


- Greg Mann


On Feb. 26, 2020, 12:12 p.m., Qian Zhang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/71886/
> -----------------------------------------------------------
> 
> (Updated Feb. 26, 2020, 12:12 p.m.)
> 
> 
> Review request for mesos, Andrei Budnik and Greg Mann.
> 
> 
> Bugs: MESOS-10047
>     https://issues.apache.org/jira/browse/MESOS-10047
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Set container's `cpu.cfs_quota_us` to its CPU resource limit.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/mesos/isolators/cgroups/subsystems/cpu.cpp 960bd141430387e076a8fab1948d07719613ed90

> 
> 
> Diff: https://reviews.apache.org/r/71886/diff/4/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Qian Zhang
> 
>


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