On Fri, Apr 01, 2022 at 02:11:51AM -0700, Yosry Ahmed wrote: > On Thu, Mar 31, 2022 at 10:25 AM Roman Gushchin > <roman.gushchin@xxxxxxxxx> wrote: > > > > On Thu, Mar 31, 2022 at 08:41:51AM +0000, Yosry Ahmed wrote: > > > From: Shakeel Butt <shakeelb@xxxxxxxxxx> > > > > > > Introduce an memcg interface to trigger memory reclaim on a memory cgroup. > > > > > > Use case: Proactive Reclaim > > > --------------------------- > > > > > > A userspace proactive reclaimer can continuously probe the memcg to > > > reclaim a small amount of memory. This gives more accurate and > > > up-to-date workingset estimation as the LRUs are continuously > > > sorted and can potentially provide more deterministic memory > > > overcommit behavior. The memory overcommit controller can provide > > > more proactive response to the changing behavior of the running > > > applications instead of being reactive. > > > > > > A userspace reclaimer's purpose in this case is not a complete replacement > > > for kswapd or direct reclaim, it is to proactively identify memory savings > > > opportunities and reclaim some amount of cold pages set by the policy > > > to free up the memory for more demanding jobs or scheduling new jobs. > > > > > > A user space proactive reclaimer is used in Google data centers. > > > Additionally, Meta's TMO paper recently referenced a very similar > > > interface used for user space proactive reclaim: > > > https://dl.acm.org/doi/pdf/10.1145/3503222.3507731 > > > > > > Benefits of a user space reclaimer: > > > ----------------------------------- > > > > > > 1) More flexible on who should be charged for the cpu of the memory > > > reclaim. For proactive reclaim, it makes more sense to be centralized. > > > > > > 2) More flexible on dedicating the resources (like cpu). The memory > > > overcommit controller can balance the cost between the cpu usage and > > > the memory reclaimed. > > > > > > 3) Provides a way to the applications to keep their LRUs sorted, so, > > > under memory pressure better reclaim candidates are selected. This also > > > gives more accurate and uptodate notion of working set for an > > > application. > > > > > > Why memory.high is not enough? > > > ------------------------------ > > > > > > - memory.high can be used to trigger reclaim in a memcg and can > > > potentially be used for proactive reclaim. > > > However there is a big downside in using memory.high. It can potentially > > > introduce high reclaim stalls in the target application as the > > > allocations from the processes or the threads of the application can hit > > > the temporary memory.high limit. > > > > > > - Userspace proactive reclaimers usually use feedback loops to decide > > > how much memory to proactively reclaim from a workload. The metrics > > > used for this are usually either refaults or PSI, and these metrics > > > will become messy if the application gets throttled by hitting the > > > high limit. > > > > > > - memory.high is a stateful interface, if the userspace proactive > > > reclaimer crashes for any reason while triggering reclaim it can leave > > > the application in a bad state. > > > > > > - If a workload is rapidly expanding, setting memory.high to proactively > > > reclaim memory can result in actually reclaiming more memory than > > > intended. > > > > > > The benefits of such interface and shortcomings of existing interface > > > were further discussed in this RFC thread: > > > https://lore.kernel.org/linux-mm/5df21376-7dd1-bf81-8414-32a73cea45dd@xxxxxxxxxx/ > > > > Hello! > > > > I'm totally up for the proposed feature! It makes total sense and is proved > > to be useful, let's add it. > > > > > > > > Interface: > > > ---------- > > > > > > Introducing a very simple memcg interface 'echo 10M > memory.reclaim' to > > > trigger reclaim in the target memory cgroup. > > > > > > > > > Possible Extensions: > > > -------------------- > > > > > > - This interface can be extended with an additional parameter or flags > > > to allow specifying one or more types of memory to reclaim from (e.g. > > > file, anon, ..). > > > > > > - The interface can also be extended with a node mask to reclaim from > > > specific nodes. This has use cases for reclaim-based demotion in memory > > > tiering systens. > > > > > > - A similar per-node interface can also be added to support proactive > > > reclaim and reclaim-based demotion in systems without memcg. > > > > Maybe an option to specify a timeout? That might simplify the userspace part. > > Also, please please add a test to selftests/cgroup/memcg tests. > > It will also provide an example on how the userspace can use the feature. > > > > Hi Roman, thanks for taking the time to review this! > > A timeout can be a good extension, I will add it to the commit message > in the next version in possible extensions. > > I will add a test in v2, thanks! Great, thank you! > > > > > > > > > For now, let's keep things simple by adding the basic functionality. > > > > What I'm worried about is how we gonna extend it? How do you see the interface > > with 2-3 extensions from the list above? All these extensions look very > > reasonable to me, so we'll likely have to implement them soon. So let's think > > about the extensibility now. > > > > My idea is to have these extensions as optional positional arguments > (like Wei suggested), so that the interface does not get too > complicated for users who don't care about tuning these options. If > this is the case then I think there is nothing to worry about. > Otherwise, if you think some of these options make sense to be a > required argument instead, we can rethink the initial interface. The interface you're proposing is not really extensible, so we'll likely need to introduce a new interface like memory.reclaim_ext very soon. Why not create an extensible API from scratch? I'm looking at cgroup v2 documentation which describes various interface files formats and it seems like given the number of potential optional arguments the best option is nested keyed (please, refer to the Interface Files section). E.g. the format can be: echo "1G type=file nodemask=1-2 timeout=30s" > memory.reclaim We can say that now we don't support any keyed arguments, but they can be added in the future. Basically you don't even need to change any code, only document the interface properly, so we can extend it later without breaking the API. > > > I wonder if it makes more sense to introduce a sys_reclaim() syscall instead? > > In the end, such a feature might make sense on the system level too. > > Yes, there is the drop_caches sysctl, but it's too radical for many cases. > > > > I think in the RFC discussion there was consensus to add both a > per-memcg knob, as well as per-node / per-system knobs (through sysfs > or syscalls) later. Wei also points out that it's not common for a > syscall to have a cgroup argument. Actually there are examples (e.g. sys_bpf), but my only point is to make the API extensible, so maybe syscall is not the best idea. I'd add the root level interface from scratch: the code change is simple and it makes sense as a feature. Then likely we don't really need another system-level interface at all. Thanks!
