Monday, 2026-06-15

opendevreviewwanghongtao proposed openstack/nova master: Fix operator precedence in limit_check  https://review.opendev.org/c/openstack/nova/+/99329706:51
opendevreviewShalini Srivastava proposed openstack/nova master: Filter invalid kwargs in volume_api.create to prevent TypeError  https://review.opendev.org/c/openstack/nova/+/99330607:26
opendevreviewKamil Sambor proposed openstack/nova master: Restore 'fork' start method for daemon mode on Python 3.14  https://review.opendev.org/c/openstack/nova/+/98781507:39
opendevreviewAshish Gupta proposed openstack/placement master: tests: Add connection parameter to Database and PlacementFixture  https://review.opendev.org/c/openstack/placement/+/99310607:57
opendevreviewStephen Finucane proposed openstack/nova master: docs: Speed up release notes builds  https://review.opendev.org/c/openstack/nova/+/98921209:17
opendevreviewStephen Finucane proposed openstack/nova master: docs: Speed up release notes builds  https://review.opendev.org/c/openstack/nova/+/98921209:21
opendevreviewJoan Gilabert proposed openstack/nova master: Add mtty/mdpy support for testing fake mdevs  https://review.opendev.org/c/openstack/nova/+/89810009:35
opendevreviewJoan Gilabert proposed openstack/nova master: Rename vtpm job and add mtty support for vpgu test  https://review.opendev.org/c/openstack/nova/+/92214009:35
opendevreviewShalini Srivastava proposed openstack/nova master: Filter invalid kwargs in volume_api.create to prevent TypeError  https://review.opendev.org/c/openstack/nova/+/99330610:32
opendevreviewLajos Katona proposed openstack/nova master: Use SDK for Neutron Ports  https://review.opendev.org/c/openstack/nova/+/96929811:17
opendevreviewLajos Katona proposed openstack/nova master: Use SDK for Neutron security-groups  https://review.opendev.org/c/openstack/nova/+/98114111:19
opendevreviewJoan Gilabert proposed openstack/nova master: Rename vtpm job and add mtty support for vpgu test  https://review.opendev.org/c/openstack/nova/+/92214011:33
bauzassean-k-mooney[m]: gibi; could either of you look at https://review.opendev.org/q/topic:%22bug/2153425%22 ? I'm all good 12:25
opendevreviewJoan Gilabert proposed openstack/nova master: Rename vtpm job and add mtty support for vpgu test  https://review.opendev.org/c/openstack/nova/+/92214012:28
-opendevstatus- NOTICE: Recent POST_FAILURE job results with no logs were due to upload errors in one of our providers, which has been temporarily disabled now so rechecking those should be safe12:44
sean-k-mooneybauzas: sorry was on a call but i can take a look quickly12:45
sean-k-mooneybauzas: so i have been watching the updates on this passivly12:46
sean-k-mooneybauzas: this is not quite enough to fix numa vswitch correct?12:46
sean-k-mooneyits related but this is only for the live migration path12:47
bauzasit should be good12:47
sean-k-mooneywe still have the issue with the compute not havig it on spwawn12:47
bauzasat least we have a customer asking for it12:47
sean-k-mooneyor cold migrate12:47
sean-k-mooneybauzas: what i mean is this will help with live migration but there is more work needed to make sure that all code path include the limits12:48
bauzaswe can ask that for a follow-up12:49
sean-k-mooneybugs.launchpad.net/nova/+bug/2145135 we alos have a gap for https://bugs.launchpad.net/nova/+bug/185533212:49
sean-k-mooneybauzas: sure im not saying its inscope fo those change12:49
sean-k-mooneybut you were previsoly working on bugs.launchpad.net/nova/+bug/2145135 a month or two ago12:49
sean-k-mooneythat why im bringing the topic up12:50
sean-k-mooneyanyway ill review it now i just want to clear on the scope12:50
bauzasack thanks12:52
sean-k-mooneyjust an fyi the expecation in the bug are also incorrect  https://bugs.launchpad.net/nova/+bug/215342512:53
sean-k-mooneysetting only  `{‘hw:numa_nodes’: ‘1’}`12:53
sean-k-mooneydoes not claim any cpu or memroy in the host numa tracker but it does end up with the vm being pinned12:53
sean-k-mooneyto a host numa cell12:53
sean-k-mooneywe do not expect any numa blancing to happen with that config unless you also request hw:mem_page_size in teh falvor or image12:54
sean-k-mooneybauzas: so im sorry to say but thet bug is technially inviald and its a psudo feature request.12:58
bauzasbut the issue still remains 13:00
sean-k-mooneythe flavor is not valid13:00
sean-k-mooneywe can add supprot for numa blancin on shared cpus only but that was never supproted by nova13:00
sean-k-mooneywe supprot it for pinned cpus13:00
sean-k-mooneyand for guest memory13:00
sean-k-mooneybut not for shared cpus13:01
sean-k-mooneyso im not agains supproting that13:01
sean-k-mooneybut the expecation today is that we wont consider the cpu allcation ration at the numa node level today13:01
bauzashonestly, I don't know what to say :(13:09
sean-k-mooneybauzas: im not agaisnt building thei fucntionlatiy into nova but its really a new feature13:10
sean-k-mooneyand if we are to do it we need to do it for spawn and cold migrate and unshleve ectra13:11
sean-k-mooneythisis not the first tirm that i have had this converatoion with operators or custoemrs and pointed out that simply requesting hw:numa_node and nothing else is only vaild if your usign file backed memory as we do not do numa aware shared cpu placement so we will not blance between numa node13:12
sean-k-mooneyto be clear even with file backed memory it not a good idea13:13
sean-k-mooneyif other really want to tack this as a bug we need to add supprot for this on spwan and other move operaiotn and more fucntional tests to cover that13:13
sean-k-mooneyhttps://bugs.launchpad.net/nova/+bug/1792985 and https://bugs.launchpad.net/nova/+bug/1439247 are relevent context for why the current flavor is invlaid13:23
sean-k-mooneyusign file backed memory prevetn the OOM kill but without also susign cpu pinnign it does nto adress the cpu blancing part13:24
sean-k-mooneywhich is what ye were tryign to address with that new bug13:24
*** sambork_ is now known as sambork13:32
*** iurygregory_ is now known as iurygregory13:46
opendevreviewJoan Gilabert proposed openstack/nova master: Rename vtpm job and add mtty support for vpgu test  https://review.opendev.org/c/openstack/nova/+/92214013:50
opendevreviewMerged openstack/nova stable/2025.1: Add reproducer test for bug 2105896  https://review.opendev.org/c/openstack/nova/+/98953514:00
*** ralonsoh is now known as ralonsoh_ooo15:09
dansmithsean-k-mooney: unrelated to the numa aspect, why is it not right to set self.limits before we call that compute-via-RPC check? it seems that the code expects limits to be set,.15:26
dansmithbut I'm not sure when else they would be, but I'm sure I'm missing something15:26
sean-k-mooneydansmith: that part isned nessarly wrong but the expecation that nova shoudl blance based on that is. also nova shoudl not be multiplying the core or memory on a given numa node by the ratios15:27
Ugglareminder: upstream meeting in ~30mn15:27
sean-k-mooneywell the memofy shoudl not be multipled the cores could be viable15:28
dansmithsean-k-mooney: yeah okay I was going to say.. the assumption in the patch being wrong makes sense, not arguing with that15:29
dansmithit just looks to me like in the force case we probably need to be setting self.limits15:30
sean-k-mooneydansmith: so there is a seprate bug in teh conductor15:30
sean-k-mooneywhere the limt are nto passed to the compute15:30
dansmithyeah15:30
sean-k-mooneyso the cshcdule does nto actully select the placmenet on the host15:30
sean-k-mooneythe compute does15:30
sean-k-mooneyso untile we etenbr recreate them there ro pass them15:31
sean-k-mooneythe two could still disagree15:31
dansmith"select the placement" meaning the numa placement?15:31
dansmithwe're still calling scheduler even in the force case right?15:31
sean-k-mooneyif you mean force live migration i belive if you use the old api where that exist15:32
sean-k-mooneywe only check that the host exist and skipp all the filters15:32
sean-k-mooneybut if you use the non force microverion where you can pass a host15:32
sean-k-mooneythen we run true the filters15:32
dansmithsean-k-mooney: L100 here https://review.opendev.org/c/openstack/nova/+/990212/5/nova/conductor/tasks/live_migrate.py#15:33
sean-k-mooneyso if you use the old foce api then we just trust you that it fits and send it to the compute15:33
dansmithoh I guess in that case we don't even have limits to set15:33
sean-k-mooneyright because we are currently relying on a sideffect fo the numa topltoy filter to create them15:34
dansmithyeah, okay15:34
sean-k-mooneywhich is the bug with numa vswthces15:34
sean-k-mooneywithout checking i woudl have to assume we are effectivly applying our defaults or no multipler at all when the limits are not pass today?15:36
sean-k-mooneywhich while not ideal for memory and disk its at least <=1.0 now15:37
bauzassean-k-mooney: the problem is that we say 'no' to something is a tribal knowledge miss15:37
sean-k-mooneyin practice we are relying on placment to enfoce the multipler globally on the compute node15:37
bauzassean-k-mooney: if the scheduler provides limits and then the conductor doesn't use it, then it's a bug15:38
sean-k-mooneybauzas: this is somethign we have discussed about changign a few time and when i have suggested it we wer econcerd about upgrade impact15:38
bauzasbut I understand your point, the scheduler shouldn't accept this15:38
sean-k-mooneybauzas: dansmith  just to be clear15:38
sean-k-mooneyim not agaisnt fixign the fact we shoudl pass limit when we dont15:38
sean-k-mooneybut im not ok with impliy hw:numa_nodes=1 with nothign else is a vlaid numa flavor that shoudl be balanced15:39
sean-k-mooneyat elast not until that works for all code paths15:39
bauzasso we should provide a HTTP400 if an operator creates a wrong flavor then15:39
bauzasat least when creating the instance15:39
sean-k-mooneywe coudl btu we rejected that in the past15:40
sean-k-mooneywe dicuss this when stephen was addign the falvor metadata validation15:40
sean-k-mooneythe minitum valid numa flavor to day has hw:mem_page_size set to anything15:40
bauzasbut that's a tribal knowledge, right?15:41
sean-k-mooneyeverythign else builds form that baseline otherwise your vm will get OOM killed eventually15:41
sean-k-mooneybauzas: sort of 15:41
bauzasbecause I wonder why we should just use this flavor extraspec (mem_page_size) just for NUMA usage ?15:41
sean-k-mooneybecause without it we do non numa aware memroy tracking15:41
sean-k-mooneyand we pin you to a numa ndoe without chekcign if the sum of vms on that numa ndoe fit15:42
bauzasso that's a tech debt15:42
sean-k-mooneyyep from day one of numa supprot in tree15:42
bauzasoperators should be able to ask for NUMA nodes without needing to ask page sizes15:42
sean-k-mooneyand its tech debt i hav eproposed fixing multipel time15:42
sean-k-mooneybut we have rejected because it breaks upgrades15:43
bauzasbut at least for the bug report, I think we can still accept it15:43
sean-k-mooneywe can accpt the bug for limits15:43
bauzasif the scheduler provides limits, we could provide them to the conductor and then the compute15:43
sean-k-mooneynot for numa without pagesize15:43
bauzasthen the scheduler shouldn't provide the limits15:44
bauzasright?15:44
sean-k-mooneysorry im not following the question15:46
bauzaslemme explain it better my question15:47
bauzasif we don't want nova to accept flavors asking numa cores with no page sizes, then we should not pass limits down to the conductor15:48
sean-k-mooneyhttps://etherpad.opendev.org/p/nova-wallaby-ptg#L71215:48
bauzasif we're passing scheduler limits to the conductor, this means "oh, yeah, I accept this instance, please check those limits when you run the claims"15:48
sean-k-mooneythat was the last time i brought up chagnitn this at the ptg15:48
bauzassorry, my brain is bad about remembering things in general :-(15:49
sean-k-mooneybauzas: the limits we are checkign are cpu ram and disk15:49
sean-k-mooneythose are enfocced by placment at the host level15:49
sean-k-mooneythe have never applied with the host at the numa level15:49
bauzaslike, I wasn't even to recollate the bug with my own numa vswitches 15:49
sean-k-mooneythat whwere the diconnect is. we have never considerd the allcation tratios when it comes to numa affinity15:50
sean-k-mooneybauzas: back in wallaby https://etherpad.opendev.org/p/r.321f34cf3eb9caa9d87a9ec8349c3d29#L712 we agreed an aproch to adress this so that any numa instance woudl be give a hw:mem_page_size if not set15:52
sean-k-mooneythat didnt happen because it was internally depriorised by our pm so i didnt end up workign on that15:52
bauzasthanks, again, I wasn't able to remember this, sorry15:53
bauzasbut now we stora kind of accepting those flavors and we can't say 'well, doh, we don't support this, thanks'15:54
sean-k-mooneyno worries. so to have a path forward, i have no issues with passing limit when they shoudl be passed, and i have no issue with closign this numa footgun15:54
bauzasI'm looking at the upstream docs to see whether we call it15:54
sean-k-mooneybauzas: ya that has been the main concurn how do we fix this without breakign upgrades15:54
bauzassean-k-mooney: so, to clarify, you say "let's provide limits when we have them, but let's not pass those limits if those shouldn't be supported". Am I correct with this assumptio ?15:55
sean-k-mooneymore or less. 15:56
sean-k-mooneyim kind fo ok wiht alwasy passign them but my main issue with the bug is expecting that nova will do the numa blancing as a side effeict15:56
sean-k-mooneyit might but that was never inteded to work15:56
bauzashmmmm15:57
bauzas(I'm trying to find how we couldn't provide the limits if we have a specific flavor)15:57
sean-k-mooneyits a littel tricky because you can change the limite via placement15:57
bauzaswell, not really 'how' but 'where'15:57
bauzasyeah15:58
sean-k-mooneyso there are 2 sepreate thing there are the numa toplogy constratis which  we can create form flavor and iamge metadta15:58
sean-k-mooneyand seperate there is the limige for the allcoation ratios15:58
opendevreviewhuangjs3 proposed openstack/nova master: cpu monitor: split zero/negative cputime paths  https://review.opendev.org/c/openstack/nova/+/99320515:59
sean-k-mooneyi think we get those in the prover summaries15:59
sean-k-mooneybut i think the metign is stating now issh?15:59
sean-k-mooneyso we can pick this up after15:59
Uggla#startmeeting nova15:59
opendevmeetMeeting started Mon Jun 15 15:59:58 2026 UTC and is due to finish in 60 minutes.  The chair is Uggla. Information about MeetBot at http://wiki.debian.org/MeetBot.15:59
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:59
opendevmeetThe meeting name has been set to 'nova'15:59
UgglaHello everyone16:00

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!