On Sat, Jul 6, 2019 at 3:54 AM Michal Hocko <mhocko@xxxxxxxxxx> wrote: > > On Fri 05-07-19 22:33:04, Yafang Shao wrote: > > On Fri, Jul 5, 2019 at 7:10 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote: > > > > > > On Fri 05-07-19 17:41:44, Yafang Shao wrote: > > > > On Fri, Jul 5, 2019 at 5:09 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote: > > > [...] > > > > > Why cannot you move over to v2 and have to stick with v1? > > > > Because the interfaces between cgroup v1 and cgroup v2 are changed too > > > > much, which is unacceptable by our customer. > > > > > > Could you be more specific about obstacles with respect to interfaces > > > please? > > > > > > > Lots of applications will be changed. > > Kubernetes, Docker and some other applications which are using cgroup v1, > > that will be a trouble, because they are not maintained by us. > > Do they actually have to change or they can simply use v2? I mean, how > many of them really do rely on having tasks in intermediate nodes or > rely on per-thread cgroups? Those should be the most visibile changes in > the interface except for control files naming. If it is purely about the > naming then it should be quite trivial to update, no? > This is not a technical issue. There're many other factors we have to consider, i.e. the cost. One simple example is in publich cloud, if we upgrade our system and force the customers to modify their user code to run on our new system, we may lost these customers. As this is not a techical issue really, I don't think we can make a clear conclusion here. > Brian has already answered the xfs part I believe. I am not really > familiar with that topic so I cannot comment anyway. > > > Do you know which companies besides facebook are using cgroup v2 in > > their product enviroment? > > I do not really know who those users are but it has been made a wider > decision that v2 is going to be a rework of a new interface and the the > v1 will be preserved and maintain for ever for backward compatibility. > If there are usecases which cannot use v2 because of some fundamental > reasons then we really want to hear about those. And if v2 really is not > usable we can think of adding features to v1 of course. > -- > Michal Hocko > SUSE Labs
