2020-02-11 08:18:57

by 王贇

[permalink] [raw]
Subject: [RFC] why can't dynamic isolation just like the static way

Hi, folks

We are dealing with isolcpus these days and try to do the isolation
dynamically.

The kernel doc lead us into the cpuset.sched_load_balance, it's fine
to achieve the dynamic isolation with it, however we got problem with
the systemd stuff.

It's keeping create cgroup with sched_load_balance enabled on default,
while the cpus are overlapped with the isolated ones, which lead into
sched domain rebuild and these cpus become non-isolated.

We're just looking forward an easy way to dynamic isolate some cpus,
just like the isolation parameter, but sched_load_balance forcing us
to dealing with the management of cgroups, we really don't get the
point in here...

Why do we have to mix the isolation with cgroups? Why not just provide
a proc entry to read cpumask and rebuild the domains?

Please let us know if there is any good reason to make the dynamic
isolation in that way, appreciated in advance :-)

Regards,
Michael Wang


2020-02-11 12:22:22

by Peter Zijlstra

[permalink] [raw]
Subject: Re: [RFC] why can't dynamic isolation just like the static way

On Tue, Feb 11, 2020 at 04:17:34PM +0800, 王贇 wrote:
> Hi, folks
>
> We are dealing with isolcpus these days and try to do the isolation
> dynamically.
>
> The kernel doc lead us into the cpuset.sched_load_balance, it's fine
> to achieve the dynamic isolation with it, however we got problem with
> the systemd stuff.

Then don't use systemd :-) Also, if systemd is the problem, why are you
bugging us?

2020-02-11 14:00:31

by Steven Rostedt

[permalink] [raw]
Subject: Re: [RFC] why can't dynamic isolation just like the static way


You forgot to include the cgroup maintainers.

-- Steve


On Tue, 11 Feb 2020 16:17:34 +0800
王贇 <[email protected]> wrote:

> Hi, folks
>
> We are dealing with isolcpus these days and try to do the isolation
> dynamically.
>
> The kernel doc lead us into the cpuset.sched_load_balance, it's fine
> to achieve the dynamic isolation with it, however we got problem with
> the systemd stuff.
>
> It's keeping create cgroup with sched_load_balance enabled on default,
> while the cpus are overlapped with the isolated ones, which lead into
> sched domain rebuild and these cpus become non-isolated.
>
> We're just looking forward an easy way to dynamic isolate some cpus,
> just like the isolation parameter, but sched_load_balance forcing us
> to dealing with the management of cgroups, we really don't get the
> point in here...
>
> Why do we have to mix the isolation with cgroups? Why not just provide
> a proc entry to read cpumask and rebuild the domains?
>
> Please let us know if there is any good reason to make the dynamic
> isolation in that way, appreciated in advance :-)
>
> Regards,
> Michael Wang

2020-02-11 14:27:10

by Steven Rostedt

[permalink] [raw]
Subject: Re: [RFC] why can't dynamic isolation just like the static way

On Tue, 11 Feb 2020 12:43:50 +0100
Peter Zijlstra <[email protected]> wrote:

> On Tue, Feb 11, 2020 at 04:17:34PM +0800, 王贇 wrote:
> > Hi, folks
> >
> > We are dealing with isolcpus these days and try to do the isolation
> > dynamically.
> >
> > The kernel doc lead us into the cpuset.sched_load_balance, it's fine
> > to achieve the dynamic isolation with it, however we got problem with
> > the systemd stuff.
>
> Then don't use systemd :-) Also, if systemd is the problem, why are you
> bugging us?

[ Background. Peter is someone that doesn't even use systemd. ;-) ]

-- Steve

2020-02-12 01:32:05

by 王贇

[permalink] [raw]
Subject: Re: [RFC] why can't dynamic isolation just like the static way



On 2020/2/11 下午7:43, Peter Zijlstra wrote:
> On Tue, Feb 11, 2020 at 04:17:34PM +0800, 王贇 wrote:
>> Hi, folks
>>
>> We are dealing with isolcpus these days and try to do the isolation
>> dynamically.
>>
>> The kernel doc lead us into the cpuset.sched_load_balance, it's fine
>> to achieve the dynamic isolation with it, however we got problem with
>> the systemd stuff.
>
> Then don't use systemd :-) Also, if systemd is the problem, why are you
> bugging us?

Well, that's... fair enough :-P

What we try to understand is why dynamic isolation is so different with
the way of static isolation, is it not good to have a simple way instead?

Regards,
Michael Wang

>

2020-02-12 01:35:57

by 王贇

[permalink] [raw]
Subject: Re: [RFC] why can't dynamic isolation just like the static way



On 2020/2/11 下午10:00, Steven Rostedt wrote:
> On Tue, 11 Feb 2020 12:43:50 +0100
> Peter Zijlstra <[email protected]> wrote:
>
>> On Tue, Feb 11, 2020 at 04:17:34PM +0800, 王贇 wrote:
>>> Hi, folks
>>>
>>> We are dealing with isolcpus these days and try to do the isolation
>>> dynamically.
>>>
>>> The kernel doc lead us into the cpuset.sched_load_balance, it's fine
>>> to achieve the dynamic isolation with it, however we got problem with
>>> the systemd stuff.
>>
>> Then don't use systemd :-) Also, if systemd is the problem, why are you
>> bugging us?
>
> [ Background. Peter is someone that doesn't even use systemd. ;-) ]

I would be happy to get rid of that too ;-) but seems like it's getting
popular now as the basic init stuff, and I guess they have no idea about
how they are breaking the dynamic isolation.

Regards,
Michael Wang

>
> -- Steve
>