On 2013/6/5 4:01, Tejun Heo wrote: > On Tue, Jun 04, 2013 at 01:21:39PM +0200, Michal Hocko wrote: >> On Mon 03-06-13 19:13:02, Tejun Heo wrote: >>> Some resources controlled by cgroup aren't per-task and cgroup core >>> allowing threads of a single thread_group to be in different cgroups >>> forced memcg do explicitly find the group leader and use it. This is >>> gonna be nasty when transitioning to unified hierarchy and in general >>> we don't want and won't support granularity finer than processes. >>> >>> Mark "tasks" with CFTYPE_INSANE. >> >> Hmm, I wasn't aware that procs is a better interface to work with >> entities in the group so I was using tasks which worked well for memcg. >> I am afraid I am not the only one. Can we get a warning when somebody >> opens the file? >> >> That being said, I do not object against removal, please just add a >> warning to let people know that procs is a preferred interface. > > Hmmm... I don't know. For users of multiple hierarchies, tasks are > fine. It's only gonna be an issue when we transition to unified > hierarchy where a lot of other things would change too. I'm not sure > whether it'd be worthwhile to generate a warning now for everyone. > Li, what do you think? > I think users who use cgroup for serious work should already know that "tasks" can be used to attach a thread only, and the "cgroup.procs" file is documented in Documentation/cgroups/cgroups.txt As we'll keep allowing multiple hierarchies for forseeable future, and we don't have a timeline to forbid it, I don't think we want to pump a warning for now. And we don't generate warnings for other "insane" behaviors currently. _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linuxfoundation.org/mailman/listinfo/containers
