Re: cgroups(7): documenting cgroups v2 delegation

[Tejun Heo]

Hello,

On Tue, Jan 09, 2018 at 12:27:58AM +0100, Michael Kerrisk (man-pages) wrote:
> >>        /dlgt_grp/cgroup.subtree_control
> >>               Making this file owned by  the  delegatee  is  optional.
> >>               Doing  so  means that that the delegatee can enable conâ
> >>               trollers  (that  are  present  in  /dlgt_grp/cgroup.conâ
> >>               trollers)  in order to further redistribute resources at
> >>               lower levels in  the  subtree.   As  an  alternative  to
> >>               changing the ownership of this file, the delegater might
> >>               instead add selected controllers to this file.
> > 
> > I'm not sure how useful it is to describe this to be optional.  In the
> > same sense, cgroup.procs would be optional too as the delegatee can
> > take control from its first children.  Users of course can choose to
> > do mix and match as they see fit but outside of the defined model,
> > there can be surprises - e.g. nsdelegate or some future delegation
> > aware feature can behave differently.  I think it'd be better to keep
> > it simple - either a subtree is delegated or not.
> 
> So, I changed that piece to:
> 
>        /dlgt_grp/cgroup.subtree_control
>               Changing the ownership of this file  means  that  that  the
>               delegatee  can  enable  controllers  (that  are  present in
>               /dlgt_grp/cgroup.controllers) in order  to  further  redisâ
>               tribute  resources  at lower levels in the subtree.  (As an
>               alternative to changing the ownership  of  this  file,  the
>               delegater  might  instead  add selected controllers to this
>               file.)
> 
> Okay?

I haven't thought much about not giving out cgroup.subtree_control
when delegating.  It's probably okay but there can be surprises -
e.g. systemd or other agent failing to init resource control because
it failed to write to subtree_control at root.

Also, given that the recommended way to delegate is now chown all
files in the /sys/kernel/cgroup/delegate file it's a bit weird to
single out subtree_control.

That said, not necessarily an objection.  I'm just not sure about it.

> > Roman recently added /sys/kernel/cgroup/delegate and
> > /sys/kernel/cgroup/features.  The former contains newline separated
> > list of cgroup files which should be chowned on delegation (in
> > addition to the directory itself) and the latter contains optional
> > features (currently only nsdelegate).  
> 
> Oh -- and this reminds that I've been meaning to ask you for a while
> now: could you please (please please please) CC all cgroup interface
> changes to linux-api@xxxxxxxxxxxxxxx (and prod others to do so
> please). There have been many of these changes in recent times
> (addition of new v2 controllers, thread mode, nsdelegate, the
> changes that Roman made that you refer to above), and they really
> all should have been CCed to linux-api@. It's often the only (easy)
> way that I have to discover changes that should be documented in
> the manual pages. And there are many other groups that are also
> interested in tracking kernel-user-space interface changes; see 
> https://www.kernel.org/doc/man-pages/linux-api-ml.html

Sorry about having not added them before.  Will try to.  Please feel
free to scream at me if I forget.

> > Also, if nsdelegate is enabled, both the source and destination
> > cgroups must be visible (cgroup namespace-wise) to the writer.
> 
> I added the following:
> 
>        *  If  the  cgroup  v2  filesystem was mounted with the nsdelegate
>           option, the writer must be able to see the source and  destinaâ
>           tion cgroup from its cgroup namespace.
> 
> Okay?

Yeah, thanks.

-- 
tejun