From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:48200) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1h0qVQ-0002Yz-3m for qemu-devel@nongnu.org; Mon, 04 Mar 2019 11:31:57 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1h0qVO-0004oa-Gu for qemu-devel@nongnu.org; Mon, 04 Mar 2019 11:31:56 -0500 Date: Mon, 4 Mar 2019 16:31:44 +0000 From: "Dr. David Alan Gilbert" Message-ID: <20190304163143.GB2596@work-vm> References: <1551454936-205218-2-git-send-email-imammedo@redhat.com> <20190301154947.GJ21251@redhat.com> <20190301183328.20b63e23@redhat.com> <20190301174806.GN21251@redhat.com> <87va0zcdse.fsf@dusky.pond.sub.org> <20190304132507.39273826@redhat.com> <20190304123908.GK4239@redhat.com> <20190304151641.3deefc3b@redhat.com> <20190304142432.GM4239@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [libvirt] [PATCH 1/2] numa: deprecate 'mem' parameter of '-numa node' option List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Michal Privoznik Cc: Daniel =?iso-8859-1?Q?P=2E_Berrang=E9?= , Igor Mammedov , Markus Armbruster , peter.maydell@linaro.org, ehabkost@redhat.com, libvir-list@redhat.com, qemu-devel@nongnu.org, qemu-arm@nongnu.org, qemu-ppc@nongnu.org, pbonzini@redhat.com, david@gibson.dropbear.id.au * Michal Privoznik (mprivozn@redhat.com) wrote: > On 3/4/19 3:24 PM, Daniel P. Berrang=E9 wrote: > > On Mon, Mar 04, 2019 at 03:16:41PM +0100, Igor Mammedov wrote: > > > On Mon, 4 Mar 2019 12:39:08 +0000 > > > Daniel P. Berrang=E9 wrote: > > >=20 > > > > On Mon, Mar 04, 2019 at 01:25:07PM +0100, Igor Mammedov wrote: > > > > > On Mon, 04 Mar 2019 08:13:53 +0100 > > > > > Markus Armbruster wrote: > > > > > > Daniel P. Berrang=E9 writes: > > > > > > > On Fri, Mar 01, 2019 at 06:33:28PM +0100, Igor Mammedov wro= te: > > > > > > > > On Fri, 1 Mar 2019 15:49:47 +0000 > > > > > > > > Daniel P. Berrang=E9 wrote: > > > > > > > > > On Fri, Mar 01, 2019 at 04:42:15PM +0100, Igor Mammedov= wrote: > > > > > > > > > > The parameter allows to configure fake NUMA topology = where guest > > > > > > > > > > VM simulates NUMA topology but not actually getting a= performance > > > > > > > > > > benefits from it. The same or better results could be= achieved > > > > > > > > > > using 'memdev' parameter. In light of that any VM tha= t uses NUMA > > > > > > > > > > to get its benefits should use 'memdev' and to allow = transition > > > > > > > > > > initial RAM to device based model, deprecate 'mem' pa= rameter as > > > > > > > > > > its ad-hoc partitioning of initial RAM MemoryRegion c= an't be > > > > > > > > > > translated to memdev based backend transparently to u= sers and in > > > > > > > > > > compatible manner (migration wise). > > > > > > > > > >=20 > > > > > > > > > > That will also allow to clean up a bit our numa code,= leaving only > > > > > > > > > > 'memdev' impl. in place and several boards that use n= ode_mem > > > > > > > > > > to generate FDT/ACPI description from it. > > > > > > > > >=20 > > > > > > > > > Can you confirm that the 'mem' and 'memdev' parameters= to -numa > > > > > > > > > are 100% live migration compatible in both directions ?= Libvirt > > > > > > > > > would need this to be the case in order to use the 'mem= dev' syntax > > > > > > > > > instead. > > > > > > > > Unfortunately they are not migration compatible in any di= rection, > > > > > > > > if it where possible to translate them to each other I'd = alias 'mem' > > > > > > > > to 'memdev' without deprecation. The former sends over on= ly one > > > > > > > > MemoryRegion to target, while the later sends over severa= l (one per > > > > > > > > memdev). > > > > > > >=20 > > > > > > > If we can't migration from one to the other, then we can no= t deprecate > > > > > > > the existing 'mem' syntax. Even if libvirt were to provide = a config > > > > > > > option to let apps opt-in to the new syntax, we need to be = able to > > > > > > > support live migration of existing running VMs indefinitely= . Effectively > > > > > > > this means we need the to keep 'mem' support forever, or at= least such > > > > > > > a long time that it effectively means forever. > > > > > > >=20 > > > > > > > So I think this patch has to be dropped & replaced with one= that > > > > > > > simply documents that memdev syntax is preferred. > > > > > >=20 > > > > > > We have this habit of postulating absolutes like "can not dep= recate" > > > > > > instead of engaging with the tradeoffs. We need to kick it. > > > > > >=20 > > > > > > So let's have an actual look at the tradeoffs. > > > > > >=20 > > > > > > We don't actually "support live migration of existing running= VMs > > > > > > indefinitely". > > > > > >=20 > > > > > > We support live migration to any newer version of QEMU that s= till > > > > > > supports the machine type. > > > > > >=20 > > > > > > We support live migration to any older version of QEMU that a= lready > > > > > > supports the machine type and all the devices the machine use= s. > > > > > >=20 > > > > > > Aside: "support" is really an honest best effort here. If yo= u rely on > > > > > > it, use a downstream that puts in the (substantial!) QA work = real > > > > > > support takes. > > > > > >=20 > > > > > > Feature deprecation is not a contract to drop the feature aft= er two > > > > > > releases, or even five. It's a formal notice that users of t= he feature > > > > > > should transition to its replacement in an orderly manner. > > > > > >=20 > > > > > > If I understand Igor correctly, all users should transition a= way from > > > > > > outdated NUMA configurations at least for new VMs in an order= ly manner. > > > > > Yes, we can postpone removing options until there are machines = type > > > > > versions that were capable to use it (unfortunate but probably > > > > > unavoidable unless there is a migration trick to make transitio= n > > > > > transparent) but that should not stop us from disabling broken > > > > > options on new machine types at least. > > > > >=20 > > > > > This series can serve as formal notice with follow up disabling= of > > > > > deprecated options for new machine types. (As Thomas noted, jus= t warnings > > > > > do not work and users continue to use broken features regardles= s whether > > > > > they are don't know about issues or aware of it [*]) > > > > >=20 > > > > > Hence suggested deprecation approach and enforced rejection of = legacy > > > > > numa options for new machine types in 2 releases so users would= stop > > > > > using them eventually. > > > >=20 > > > > When we deprecate something, we need to have a way for apps to us= e the > > > > new alternative approach *at the same time*. So even if we only = want to > > > > deprecate for new machine types, we still have to first solve the= problem > > > > of how mgmt apps will introspect QEMU to learn which machine type= s expect > > > > the new options. > > > I'm not aware any mechanism to introspect machine type options (exi= sting > > > or something being developed). Are/were there any ideas about it th= at were > > > discussed in the past? > > >=20 > > > Aside from developing a new mechanism what are alternative approach= es? > > > I mean when we delete deprecated CLI option, how it's solved on lib= virt > > > side currently? > > >=20 > > > For example I don't see anything introspection related when we have= been > > > removing deprecated options recently. > >=20 > > Right, with other stuff we deprecate we've had a simpler time, as it > > either didn't affect migration at all, or the new replacement stuff > > was fully compatible with the migration data stream. IOW, libvirt > > could unconditionally use the new feature as soon as it saw that it > > exists in QEMU. We didn't have any machine type dependancy to deal > > with before now. >=20 > We couldn't have done that. How we would migrate from older qemu? >=20 > Anyway, now that I look into this (esp. git log) I came accross: >=20 > commit f309db1f4d51009bad0d32e12efc75530b66836b > Author: Michal Privoznik > AuthorDate: Thu Dec 18 12:36:48 2014 +0100 > Commit: Michal Privoznik > CommitDate: Fri Dec 19 07:44:44 2014 +0100 >=20 > qemu: Create memory-backend-{ram,file} iff needed >=20 > Or this 7832fac84741d65e851dbdbfaf474785cbfdcf3c. We did try to generat= ed > newer cmd line but then for various reasong (e.g. avoiding triggering a= qemu > bug) we turned it off and make libvirt default to older (now deprecated= ) cmd > line. >=20 > Frankly, I don't know how to proceed. Unless qemu is fixed to allow > migration from deprecated to new cmd line (unlikely, if not impossible, > right?) then I guess the only approach we can have is that: >=20 > 1) whenever so called cold booting a new machine (fresh, brand new star= t of > a new domain) libvirt would default to modern cmd line, > > 2) on migration, libvirt would record in the migration stream (or statu= s XML > or wherever) that modern cmd line was generated and thus it'll make the > destination generate modern cmd line too. >=20 > This solution still suffers a couple of problems: > a) migration to older libvirt will fail as older libvirt won't recogniz= e the > flag set in 2) and therefore would default to deprecated cmd line > b) migrating from one host to another won't modernize the cmd line >=20 > But I guess we have to draw a line somewhere (if we are not willing to = write > those migration patches). What's interesting here is that this problem isn't really machine-type related; so providing introspection on the machine type doesn't immediately help. What we're actually trying to do here is (mis)use a machine type as a proxy for knowing that both sides are new enough to handle the new command line. That's an OK thing to do, and if we did have introspection we could add a fudge flag to say it's allowed now. Dave > Michal -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK