`On Wed, Aug 31, 2011 at 08:29:24AM +0900, KAMEZAWA Hiroyuki wrote:> On Tue, 30 Aug 2011 13:32:21 +0200> Johannes Weiner <jweiner@redhat.com> wrote:> > > On Tue, Aug 30, 2011 at 07:38:39PM +0900, KAMEZAWA Hiroyuki wrote:> > > On Tue, 30 Aug 2011 12:17:26 +0200> > > Johannes Weiner <jweiner@redhat.com> wrote:> > > > > > > On Tue, Aug 30, 2011 at 05:56:09PM +0900, KAMEZAWA Hiroyuki wrote:> > > > > On Tue, 30 Aug 2011 10:42:45 +0200> > > > > Johannes Weiner <jweiner@redhat.com> wrote:> > > > > > > > > > Assume 3 cgroups in a hierarchy.> > > > > > > > > > > > > > 	A> > > > > > >        /> > > > > > >       B> > > > > > >      /> > > > > > >     C> > > > > > > > > > > > > > C's scan contains 3 causes.> > > > > > > 	C's scan caused by limit of A.> > > > > > > 	C's scan caused by limit of B.> > > > > > > 	C's scan caused by limit of C.> > > > > > >> > > > > > > If we make hierarchy sum at read, we think> > > > > > > 	B's scan_stat = B's scan_stat + C's scan_stat> > > > > > > But in precice, this is> > > > > > > > > > > > > > 	B's scan_stat = B's scan_stat caused by B +> > > > > > > 			B's scan_stat caused by A +> > > > > > > 			C's scan_stat caused by C +> > > > > > > 			C's scan_stat caused by B +> > > > > > > 			C's scan_stat caused by A.> > > > > > > > > > > > > > In orignal version.> > > > > > > 	B's scan_stat = B's scan_stat caused by B +> > > > > > > 			C's scan_stat caused by B +> > > > > > > > > > > > > > After this patch,> > > > > > > 	B's scan_stat = B's scan_stat caused by B +> > > > > > > 			B's scan_stat caused by A +> > > > > > > 			C's scan_stat caused by C +> > > > > > > 			C's scan_stat caused by B +> > > > > > > 			C's scan_stat caused by A.> > > > > > > > > > > > > > Hmm...removing hierarchy part completely seems fine to me.> > > > > > > > > > > > I see.> > > > > > > > > > > > You want to look at A and see whether its limit was responsible for> > > > > > reclaim scans in any children.  IMO, that is asking the question> > > > > > backwards.  Instead, there is a cgroup under reclaim and one wants to> > > > > > find out the cause for that.  Not the other way round.> > > > > > > > > > > > In my original proposal I suggested differentiating reclaim caused by> > > > > > internal pressure (due to own limit) and reclaim caused by> > > > > > external/hierarchical pressure (due to limits from parents).> > > > > > > > > > > > If you want to find out why C is under reclaim, look at its reclaim> > > > > > statistics.  If the _limit numbers are high, C's limit is the problem.> > > > > > If the _hierarchical numbers are high, the problem is B, A, or> > > > > > physical memory, so you check B for _limit and _hierarchical as well,> > > > > > then move on to A.> > > > > > > > > > > > Implementing this would be as easy as passing not only the memcg to> > > > > > scan (victim) to the reclaim code, but also the memcg /causing/ the> > > > > > reclaim (root_mem):> > > > > > > > > > > > 	root_mem == victim -> account to victim as _limit> > > > > > 	root_mem != victim -> account to victim as _hierarchical> > > > > > > > > > > > This would make things much simpler and more natural, both the code> > > > > > and the way of tracking down a problem, IMO.> > > > > > > > > > hmm. I have no strong opinion.> > > > > > > > I do :-)> > > > > > > BTW,  how to calculate C's lru scan caused by A finally ?> > > > > >             A> > >            /> > >           B> > >          /> > >         C> > > > > > At scanning LRU of C because of A's limit, where stats are recorded ?> > > > > > If we record it in C, we lose where the memory pressure comes from.> > > > It's recorded in C as 'scanned due to parent'.> > > > If you want to track down where pressure comes from, you check the> > outer container, B.  If B is scanned due to internal pressure, you> > know that C's external pressure comes from B.  If B is scanned due to> > external pressure, you know that B's and C's pressure comes from A or> > the physical memory limit (the outermost container, so to speak).> > > > The containers are nested.  If C is scanned because of the limit in A,> > then this concerns B as well and B must be scanned as well as B, as> > C's usage is fully contained in B.> > > > There is not really a direct connection between C and A that is> > irrelevant to B, so I see no need to record in C which parent was the> > cause of the pressure.  Just that it was /a/ parent and not itself.> > Then you can follow the pressure up the hierarchy tree.> > > > Answer to your original question:> > > > 	C_scan_due_to_A = C_scan_external - B_scan_internal - A_scan_external> > > > I'm confused. > > If vmscan is scanning in C's LRU,> 	(memcg == root) : C_scan_internal ++> 	(memcg != root) : C_scan_external ++Yes.> Why A_scan_external exists ? It's 0 ?> > I think we can never get numbers.Kswapd/direct reclaim should probably be accounted as A_external,since A has no limit, so reclaim pressure can not be internal.On the other hand, one could see the amount of physical memory in themachine as A's limit and account global reclaim as A_internal.I think the former may be more natural.That aside, all memcgs should have the same statistics, obviously.Scripts can easily deal with counters being zero.  If items differbetween cgroups, that would suck a lot.`