On Mon 28-01-19 09:49:05, Tejun Heo wrote: > Hello, Michal. > > On Mon, Jan 28, 2019 at 06:05:26PM +0100, Michal Hocko wrote: > > Yeah, that is quite clear. But it also assumes that the hierarchy is > > pretty stable but cgroups might go away at any time. I am not saying > > that the aggregated events are not useful I am just saying that it is > > quite non-trivial to use and catch all potential corner cases. Maybe I > > It really isn't complicated and doesn't require stable subtree. > > > am overcomplicating it but one thing is quite clear to me. The existing > > semantic is really useful to watch for the reclaim behavior at the > > current level of the tree. You really do not have to care what is > > happening in the subtree when it is clear that the workload itself > > is underprovisioned etc. Considering that such a semantic already > > existis, somebody might depend on it and we likely want also aggregated > > semantic then I really do not see why to risk regressions rather than > > add a new memory.hierarchy_events and have both. > > The problem then is that most other things are hierarchical including > some fields in .events files, so if we try to add local stats and > events, there's no good way to add them. All memcg events are represented non-hierarchical AFAICS memcg_memory_event() simply accounts at the level when it happens. Or do I miss something? Or are you talking about .events files for other controllers? -- Michal Hocko SUSE Labs
