Friday, 2025-10-10

opendevreviewSteve Baker proposed openstack/nova master: Add VNC console support for the Ironic driver  https://review.opendev.org/c/openstack/nova/+/94252801:30
opendevreviewSteve Baker proposed openstack/nova master: Add VNC console support for the Ironic driver  https://review.opendev.org/c/openstack/nova/+/94252801:47
*** mhen_ is now known as mhen01:58
opendevreviewmelanie witt proposed openstack/nova master: Add hw:tpm_secret_security extra spec validation  https://review.opendev.org/c/openstack/nova/+/94019703:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: support instances with `user` secret security  https://review.opendev.org/c/openstack/nova/+/94250203:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: support instances with `host` secret security  https://review.opendev.org/c/openstack/nova/+/94179503:32
opendevreviewmelanie witt proposed openstack/nova master: Add vtpm_secret_(uuid|value) to LibvirtLiveMigrateData  https://review.opendev.org/c/openstack/nova/+/95262803:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: support live migration of `host` secret security  https://review.opendev.org/c/openstack/nova/+/94148303:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: bump service version and require it for live migration  https://review.opendev.org/c/openstack/nova/+/96205103:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: support instances with `deployment` secret security  https://review.opendev.org/c/openstack/nova/+/94202103:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: support live migration of `deployment` secret security  https://review.opendev.org/c/openstack/nova/+/92577103:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: test live migration between hosts with different security  https://review.opendev.org/c/openstack/nova/+/95262903:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: add late check for supported TPM secret security  https://review.opendev.org/c/openstack/nova/+/95697503:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: opt-in to new TPM secret security via resize  https://review.opendev.org/c/openstack/nova/+/96205203:32
opendevreviewmelanie witt proposed openstack/nova master: TPM: add documentation and reno for live migration  https://review.opendev.org/c/openstack/nova/+/96288903:32
opendevreviewmelanie witt proposed openstack/nova master: DNM vtpm tempest  https://review.opendev.org/c/openstack/nova/+/95747703:32
opendevreviewmelanie witt proposed openstack/nova master: Add handling for vTPM secret permission error  https://review.opendev.org/c/openstack/nova/+/96364803:32
bauzasgibi: sean-k-mooney: dansmith: sorry for the late ping, but I'd appreciate reviews for the image props weigher fix https://review.opendev.org/q/topic:%22bug/2125935%2207:20
jlejeunehello, I'd also appreciate some reviews on my topic https://review.opendev.org/q/topic:%22bug/2085135%22 :)07:53
jlejeuneand also on these backports please: https://review.opendev.org/q/topic:%22bug-2044235%22 :)07:53
opendevreviewSeyeong Kim proposed openstack/nova master: libvirt: Support boot_index for multiple block devices  https://review.opendev.org/c/openstack/nova/+/96366508:39
opendevreviewBalazs Gibizer proposed openstack/placement master: Prune a_c search space by invalid prefixes  https://review.opendev.org/c/openstack/placement/+/96277610:03
opendevreviewBalazs Gibizer proposed openstack/placement master: DNM: trying a proposal from the parent's review  https://review.opendev.org/c/openstack/placement/+/96367310:03
LarsErikPhi there! I'm so excited that we've no got persistent mdevs for vgpus with mdevctl. But.. with the vendor-specific vfio framework, we're not really using mdevs anymore. Is there any documentation on you can persist the vgpu type of VF over a reboot?10:07
sean-k-mooneygibi: geminis respocne was easer to parse wehen it render the notation in markdown/latex and i asks for graphs :) but i could noe put those in a gerrit comment.10:09
sean-k-mooneygibi: https://imgur.com/a/S6e658j10:12
sean-k-mooneyi got that by copying your entire test function with the prompt "why 30?" feels like a good return on the investment10:14
sean-k-mooneymodulo bugs the recursive check is basically jsut winning10:15
sean-k-mooneywhere it looses it a negligable regressions10:16
sean-k-mooneysince the prune spacess is so small at 2 and 3  providers10:16
opendevreviewMerged openstack/nova master: Fix string format specifier  https://review.opendev.org/c/openstack/nova/+/96121110:21
opendevreviewMerged openstack/nova stable/2024.1: Fix 'nova-manage image_property set' command  https://review.opendev.org/c/openstack/nova/+/96334810:21
opendevreviewMerged openstack/nova stable/2024.1: Fix case-sensitivity for metadata keys  https://review.opendev.org/c/openstack/nova/+/96135210:22
opendevreviewMerged openstack/nova master: Adds regression test for bug LP#2085135  https://review.opendev.org/c/openstack/nova/+/96336810:22
LarsErikPadding to my question.. There is no way I can tell nova which VGPU type I want to create with this new framework, like we can with mdev, no?10:56
gibisean-k-mooney: yeah I agree we don't loose much on small spaces11:03
sean-k-mooneyLarsErikP: not really you woudl have to use a tool like nvidia https://github.com/NVIDIA/mig-parted or simialr11:06
sean-k-mooneyLarsErikP: this is not really in socpe of openstack to automate for the most part11:06
sean-k-mooneyLarsErikP: nova in particalr is a static allcoation model where you as the oepratior need to prepare the devices before deploying nova and it just manages allcoting them form teh set you have prepared11:07
sean-k-mooneyLarsErikP: i have been asked to look at cyborg a little more closely this cycle to see if it coudl have some sort of role to play there11:08
sean-k-mooneybut i have not actully started doing that really but i belive the anse is we woudl have to build that if we wanted it i.e. it cant today but could11:09
sean-k-mooneyLarsErikP: my best recomendation today as i said woudl be https://github.com/NVIDIA/mig-parted11:09
LarsErikPsean-k-mooney: okok. thanks. For the first question, that's kind of what I expected =) 11:32
LarsErikPand tbh I'm fine with that not being openstacks problem. It would be nice if there was some sort of way to make nova set "current_vgpu_type" though. Much like we could/can with mdevs11:33
LarsErikPIn the meantime I guess I'll have to make someting myself. mig-parted seems nice, but we're not really using MIG atm.. hehe11:34
jlejeunegibi: hello, I've answered to your question on https://review.opendev.org/c/openstack/nova/+/93302211:52
sean-k-mooneyLarsErikP: well i think it does non MIG thing too but it was orginally created for mig11:55
sean-k-mooneyi have nto looked at it since i forst found it a few years ago11:55
sean-k-mooneybut it seam to still have new commit so i assume it has evovled over time11:56
LarsErikPah. I'll have a closer looke then :P Thanks for letting me know of its existence =)11:56
sean-k-mooneyi like the idea of having a simple yaml file that just details this and havign a single agent that just configures it on boot11:56
LarsErikPmuch like mdevctl ;)11:56
LarsErikPI'm currently writing puppet for dynamically creating a oneshot systemd-unit for this, based on info from hiera11:57
LarsErikP:p11:57
sean-k-mooneykind of but really i feell ike this would be beter as some kidn fo simple systemd service extention11:57
sean-k-mooneyor udev rule11:57
LarsErikPabsolutley11:58
sean-k-mooneykindof like https://www.redhat.com/en/blog/quadlet-podman11:58
LarsErikPoh. yeah. that would be nice11:58
sean-k-mooneyon the one hand i dont like having to ram alot of thing into systemd on the other hand it can be better then installing 50 billion one off tools form diffent venders for this kind of config12:00
LarsErikPmm12:00
opendevreviewBalazs Gibizer proposed openstack/nova master: DNM: test with placement fix  https://review.opendev.org/c/openstack/nova/+/96303512:06
opendevreviewBalazs Gibizer proposed openstack/nova master: DNM:Test with optimize_for_wide_provider_trees  https://review.opendev.org/c/openstack/nova/+/96303612:06
opendevreviewSeyeong Kim proposed openstack/nova master: libvirt: Support boot_index for multiple block devices  https://review.opendev.org/c/openstack/nova/+/96366513:01
opendevreviewSeyeong Kim proposed openstack/nova master: libvirt: Support boot_index for multiple block devices  https://review.opendev.org/c/openstack/nova/+/96366513:11
opendevreviewBogdan Dobrelya proposed openstack/nova master: Determine the device type from configuration first  https://review.opendev.org/c/openstack/nova/+/96370213:11
opendevreviewSeyeong Kim proposed openstack/nova master: libvirt: Support boot_index for multiple block devices  https://review.opendev.org/c/openstack/nova/+/96366513:26
LarsErikPsean-k-mooney: hehe https://paste.openstack.org/show/bFie9CvEeQyFLHNsO6Tw/13:57
LarsErikPworks :P13:57
LarsErikPfeels.. bodgy. but hey, if it works.. :P13:58
sean-k-mooneyLarsErikP: that actully a clearner implemation of tha tthen i have seen other doo14:03
sean-k-mooneyyou could add it to our docs if you like14:03
sean-k-mooneyLarsErikP: we have similar docs like this https://docs.openstack.org/nova/latest/admin/virtual-gpu.html#enable-gpu-types-compute14:04
sean-k-mooneyLarsErikP: that is how we document you shoudl enable the vfs in teh first place you caould add a secont for configurign the vgpus type on those vfs14:06
sean-k-mooneythe current example is mianly for the mdev based approch14:06
LarsErikPsean-k-mooney: oh, thanks <314:06
sean-k-mooneyso you woul proably want to add a new section to the docs for the new managed mode passthough14:07
LarsErikPthat page should probably contain a section for the vendor-specific vfio type as well14:07
LarsErikPI can look into it on monday. How do i contribute? :P14:07
sean-k-mooneyhttps://docs.openstack.org/nova/latest/admin/pci-passthrough.html#support-for-multiple-types-of-vfs14:08
sean-k-mooneythis is the ohter place you may want to coder 14:08
sean-k-mooneyLarsErikP: well the normal way woudl be to submit a patch via gerrit to update the rst file14:08
sean-k-mooneyyou coudl also file a bug for the lack fo docs for this and put the suggestins in the launchpad bug14:09
sean-k-mooneyLarsErikP: since your more intersted in the new way of doign it with vfio variant dirvers https://docs.openstack.org/nova/latest/admin/pci-passthrough.html#support-for-multiple-types-of-vfs is more correct14:09
sean-k-mooneyLarsErikP: for the mdev based approch we do have some docs on how to discove rthe relvent types https://docs.openstack.org/nova/latest/admin/virtual-gpu.html#how-to-discover-a-gpu-type14:11
sean-k-mooneyso we could add somethign similar fo the new vfio-variant driver way too if you knwo how14:11
*** amorin_ is now known as amorin14:15
LarsErikPI've learned quite a bit about this the last few days, so I'll try to contribute to the docs next week =)14:36
LarsErikPone way or another ;)14:36
LarsErikPimho, it would be helpful with some kind of "suggestion" about the migration path here as well.. I don't know if that is something that belongs in the docs14:38
LarsErikPour story here is going from Caracal on 22.04 to Epoxy on 24.04. And when you have everything setup for VGPUs with mdev on 22.04, the migration and upgrade is getting quite involved with these changes14:40
LarsErikPWe had to leave our gpu-nodes on 22.04 because of these changes as well, since the combination of 24.04/Caracal/GPUs with SR-IOV simply is not supported14:41
LarsErikPSo our migration plan would be: Resize all GPU-VMs to non-gpu flavors. Reinstall gpu-nodes and reconfigure everything that's needed. Resize the VMs back to their original (modified) GPU-flavors14:43
bauzasLarsErikP: replying late but fwiw, Epoxy supports 22.04 https://governance.openstack.org/tc/reference/runtimes/2025.1.html15:44
sean-k-mooneyLarsErikP: the migration path is unfortunatly to do a resize16:14
sean-k-mooneyso ya you kind of need to drain 1 or two hosts and move vms between then via resize16:15
sean-k-mooneyinstead of modifyign the flavor however i would stongly advice that you just keep the new and old flaovrs and eventually retire the old16:16
sean-k-mooneythat gets a littel tricky if you have heat inovled16:16
LarsErikPbauzas: but there is no UCA for epoxy in 22.04?17:14
LarsErikPsean-k-mooney: yeah, but we can live with that in our env =)17:18
opendevreviewMerged openstack/nova stable/2025.1: [PCI tracker]Remove non configured devs when freed  https://review.opendev.org/c/openstack/nova/+/96024217:54
opendevreviewMerged openstack/nova stable/2025.1: [pci]Keep used dev in Placement regardless of dev_spec  https://review.opendev.org/c/openstack/nova/+/96024318:29
opendevreviewMerged openstack/nova master: Add a regression test for ImagePropsWeigher  https://review.opendev.org/c/openstack/nova/+/96335718:29

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!