Re: [libvirt][RFC PATCH] add a new 'default' option for attribute mode in numatune

On 8/19/2020 11:24 PM, Martin Kletzander wrote: > On Tue, Aug 18, 2020 at 07:49:30AM +0000, Zang, Rui wrote: >> >> >>> -----Original Message----- >>> From: Martin Kletzander <mkletzan(a)redhat.com&gt; >>> Sent: Monday, August 17, 2020 4:58 PM >>> To: Zhong, Luyao <luyao.zhong(a)intel.com&gt; >>> Cc: libvir-list(a)redhat.com; Zang, Rui <rui.zang(a)intel.com>; Michal >>> Privoznik >>> <mprivozn(a)redhat.com&gt; >>> Subject: Re: [libvirt][RFC PATCH] add a new 'default' option for >>> attribute mode >>> in numatune >>> >>> On Tue, Aug 11, 2020 at 04:39:42PM +0800, Zhong, Luyao wrote: >>> > >>> > >>> >On 8/7/2020 4:24 PM, Martin Kletzander wrote: >>> >> On Fri, Aug 07, 2020 at 01:27:59PM +0800, Zhong, Luyao wrote: >>> >>> >>> >>> >>> >>> On 8/3/2020 7:00 PM, Martin Kletzander wrote: >>> >>>> On Mon, Aug 03, 2020 at 05:31:56PM +0800, Luyao Zhong wrote: >>> >>>>> Hi Libvirt experts, >>> >>>>> >>> >>>>> I would like enhence the numatune snippet configuration. Given a >>> >>>>> example snippet: >>> >>>>> >>> >>>>> <domain> >>> >>>>>  ... >>> >>>>>  <numatune> >>> >>>>>    <memory mode="strict" nodeset="1-4,^3"/>   >>> >>>>> <memnode cellid="0" mode="strict" nodeset="1"/>    <memnode >>> >>>>> cellid="2" mode="preferred" nodeset="2"/>  </numatune>  ... >>> >>>>> </domain> >>> >>>>> >>> >>>>> Currently, attribute mode is either 'interleave', 'strict', or >>> >>>>> 'preferred', I propose to add a new 'default'  option. I give >>> >>>>> the reason as following. >>> >>>>> >>> >>>>> Presume we are using cgroups v1, Libvirt sets cpuset.mems for all >>> >>>>> vcpu threads according to 'nodeset' in memory element. And >>> >>>>> translate the memnode element to qemu config options (--object >>> >>>>> memory-backend-ram) for per numa cell, which invoking mbind() >>> >>>>> system call at the end.[1] >>> >>>>> >>> >>>>> But what if we want using default memory policy and request each >>> >>>>> guest numa cell pinned to different host memory nodes? We can't >>> >>>>> use mbind via qemu config options, because (I quoto here) "For >>> >>>>> MPOL_DEFAULT, the nodemask and maxnode arguments must be specify >>> >>>>> the empty set of nodes." [2] >>> >>>>> >>> >>>>> So my solution is introducing a new 'default' option for attribute >>> >>>>> mode. e.g. >>> >>>>> >>> >>>>> <domain> >>> >>>>>  ... >>> >>>>>  <numatune> >>> >>>>>    <memory mode="default" nodeset="1-2"/>    <memnode >>> >>>>> cellid="0" mode="default" nodeset="1"/>    <memnode >>> >>>>> cellid="1" mode="default" nodeset="2"/>  </numatune>  ... >>> >>>>> </domain> >>> >>>>> >>> >>>>> If the mode is 'default', libvirt should avoid generating qemu >>> >>>>> command line '--object memory-backend-ram', and invokes cgroups to >>> >>>>> set cpuset.mems for per guest numa combining with numa topology >>> >>>>> config. Presume the numa topology is : >>> >>>>> >>> >>>>> <cpu> >>> >>>>>  ... >>> >>>>>  <numa> >>> >>>>>    <cell id='0' cpus='0-3' memory='512000' unit='KiB' />  >>> >>>>>   <cell id='1' cpus='4-7' memory='512000' unit='KiB' />  >>> >>>>> </numa>  ... >>> >>>>> </cpu> >>> >>>>> >>> >>>>> Then libvirt should set cpuset.mems to '1' for vcpus 0-3, and '2' >>> >>>>> for vcpus 4-7. >>> >>>>> >>> >>>>> >>> >>>>> Is this reasonable and feasible? Welcome any comments. >>> >>>>> >>> >>>> >>> >>>> There are couple of problems here.  The memory is not (always) >>> >>>> allocated by the vCPU threads.  I also remember it to not be >>> >>>> allocated by the process, but in KVM in a way that was not affected >>> >>>> by the cgroup settings. >>> >>> >>> >>> Thanks for your reply. Maybe I don't get what you mean, could you >>> >>> give me more context? But what I proposed will have no effect on >>> >>> other memory allocation. >>> >>> >>> >> >>> >> Check how cgroups work.  We can set the memory nodes that a process >>> >> will allocate from.  However to set the node for the process >>> >> (thread) QEMU needs to be started with the vCPU threads already >>> >> spawned (albeit stopped).  And for that QEMU already allocates some >>> >> memory.  Moreover if extra memory was allocated after we set the >>> >> cpuset.mems it is not guaranteed that it will be allocated by the >>> >> vCPU in that NUMA cell, it might be done in the emulator instead or >>> >> the KVM module in the kernel in which case it might not be accounted >>> >> for the process actually causing the allocation (as we've already >>> >> seen with Linux).  In all these cases cgroups will not do what you >>> >> want them to do.  The last case might be fixed, the first ones are >>> >> by default not going to work. >>> >> >>> >>>> That might be >>> >>>> fixed now, >>> >>>> however. >>> >>>> >>> >>>> But basically what we have against is all the reasons why we >>> >>>> started using QEMU's command line arguments for all that. >>> >>>> >>> >>> I'm not proposing use QEMU's command line arguments, on contrary I >>> >>> want using cgroups setting to support a new config/requirement. I >>> >>> give a solution about if we require default memory policy and memory >>> >>> numa pinning. >>> >>> >>> >> >>> >> And I'm suggesting you look at the commit log to see why we *had* to >>> >> add these command line arguments, even though I think I managed to >>> >> describe most of them above already (except for one that _might_ >>> >> already be fixed in the kernel).  I understand the git log is huge >>> >> and the code around NUMA memory allocation was changing a lot, so I >>> >> hope my explanation will be enough. >>> >> >>> >Thank you for detailed explanation, I think I get it now. We can't >>> >guarantee memory allocation matching requirement since there is a time >>> >slot before setting cpuset.mems. >>> > >>> >>> That's one of the things, although this one could be avoided (by >>> setting a global >>> cgroup before exec()). >>> >>> >>> Thanks, >>> >>> Luyao >>> >>>> Sorry, but I think it will more likely break rather than fix stuff. >>> >>>> Maybe this >>> >>>> could be dealt with by a switch in `qemu.conf` with a huge warning >>> >>>> above it. >>> >>>> >>> >>> I'm not trying to fix something, I propose how to support a new >>> >>> requirement just like I stated above. >>> >>> >>> >> >>> >> I guess we should take a couple of steps back, I don't get what you >>> >> are trying to achieve.  Maybe if you describe your use case it will >>> >> be easier to reach a conclusion. >>> >> >>> >Yeah, I do have a usecase I didn't mention before. It's a feature in >>> >kernel but not merged yet, we call it memory tiering. >>> >(https://lwn.net/Articles/802544/) >>> > >>> >If memory tiering is enabled on host, DRAM is top tier memory, and >>> >PMEM(persistent memory) is second tier memory, PMEM is shown as numa >>> >node without cpu. For short, pages can be migrated between DRAM and >>> >PMEM based on DRAM pressure and how cold/hot they are. >>> > >>> >We could configure multiple memory migrating path. For example, node 0: >>> >DRAM, node 1: DRAM, node 2: PMEM, node 3: PMEM we can make 0+2 to a >>> >group, and 1+3 to a group. In each group, page is allowed to migrated >>> >down(demotion) and up(promotion). >>> > >>> >If **we want our VMs utilizing memory tiering and with NUMA topology**, >>> >we need handle the guest memory mapping to host memory, that means we >>> >need bind each guest numa node to a memory nodes group(DRAM node + >>> PMEM >>> >node) on host. For example, guest node 0 -> host node 0+2. >>> > >>> >However, only cgroups setting can make the memory tiering work, if we >>> >use mbind() system call, demoted pages will never go back to DRAM. >>> >That's why I propose to add 'default' option and bypass mbind in QEMU. >>> > >>> >I hope I make myself understandable. I'll appreciate if you could give >>> >some suggestion. >>> > >>> >>> This comes around every couple of months/years and bites us in the >>> back no >>> matter what way we go (every time there is someone who wants it the >>> other >>> way). >>> That's why I think there could be a way for the user to specify >>> whether they will >>> likely move the memory or not and based on that we would specify `host- >>> nodes` and `policy` to qemu or not.  I think I even suggested this >>> before (or >>> probably delegated it to someone else for a suggestion so that there >>> is more >>> discussion), but nobody really replied. >>> >>> So what we need, I think, is a way for someone to set a per-domain >>> information >>> whether we should bind the memory to nodes in a changeable fashion or >>> not. >>> I'd like to have it in as well.  The way we need to do that is, >>> probably, per- >>> domain, because adding yet another switch for each place in the XML >>> where we >>> can select a NUMA memory binding would be a suicide.  There should >>> also be >>> no need for this to be enabled per memory-(module, node), so it >>> should work >>> fine. >>> >> >> Thanks for letting us know your vision about this. >> From what I understood, the "changeable fashion" means that the guest >> numa >> cell binding can be changed out of band after initial binding, either >> by system >> admin or the operating system (memory tiering in our case), or >> whatever the >> third party is.  Is that perception correct? > > Yes.  If the user wants to have the possibility of changing the binding, > then we > use *only* cgroups.  Otherwise we use the qemu parameters that will make > qemu > call mbind() (as that has other pros mentioned above).  The other option > would > be extra communication between QEMU and libvirt during start to let us > know when > to set what cgroups etc., but I don't think that's worth it. > >> It seems to me mbind() or set_mempolicy() system calls do not offer that >> flexibility of changing afterwards. So in case of QEMU/KVM, I can only >> think >> of cgroups. >> So to be specific, if we had this additional "memory_binding_changeable" >> option specified, we will try to do the guest numa constraining via >> cgroups >> whenever possible. There will probably also be conflicts in options or >> things >> that cgroups can not do. For such cases we'd fail the domain. > > Basically we'll do what we're doing now and skip the qemu `host-nodes` and > `policy` parameters with the new option.  And of course we can fail with > a nice > error message if someone wants to move the memory without the option > selected > and so on. Thanks for your comments. I'd like get it more clear about defining the interface in domain xml, then I could go into the implementation further. As you mentioned, per-domain option will be better than per-node. I go through the libvirt doamin format to look for a proper position to place this option. Then I'm thinking we could still utilizing numatune element to configure. <numatune> <memory mode="strict" nodeset="1-4,^3"/> <memnode cellid="0" mode="strict" nodeset="1"/> <memnode cellid="2" mode="preferred" nodeset="2"/> </numatune> coincidentally, the optional memory element specifies how to allocate memory for the domain process on a NUMA host. So can we utilizing this element, and introducing a new mode like "changeable" or whatever? Do you have a better name?

If the memory mode is set to 'changeable', we could ignore the mode setting for each memnode, and then we only configure by cgroups. I have not diven into code for now, expecting it could work.

Thanks, Luyao > >> If you agree with the direction, I think we can dig deeper to see what >> will >> come out. >> >> Regards, >> Zang, Rui >> >> >>> Ideally we'd discuss it with others, but I think I am only one of a >>> few people >>> who dealt with issues in this regard.  Maybe Michal (Cc'd) also dealt >>> with some >>> things related to the binding, so maybe he can chime in. >>> >>> >regards, >>> >Luyao >>> > >>> >>>> Have a nice day, >>> >>>> Martin >>> >>>> >>> >>>>> Regards, >>> >>>>> Luyao >>> >>>>> >>> >>>>> >>> [1]https://github.com/qemu/qemu/blob/f2a1cf9180f63e88bb38ff21c169d >>> >>>>> a97c3f2bad5/backends/hostmem.c#L379 >>> >>>>> >>> >>>>> >>> >>>>> [2]https://man7.org/linux/man-pages/man2/mbind.2.html >>> >>>>> >>> >>>>> -- >>> >>>>> 2.25.1 >>> >>>>> >>> >>> >>> >