Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-10-09
15:13:42 sean-k-mooney the other way that people do this without a trait
15:13:46 CeeMac so the vmx flag should automatically be exposed to the guest if KVM is configured for nested?
15:14:05 sean-k-mooney is to use the instance extra and host aggrates
15:14:20 sean-k-mooney am yes if you use host model or host-passthough as the mode
15:15:07 melwitt dansmith: I found a small bug in the cherry-pick hash check script, if you wouldn't mind reviewing https://review.opendev.org/756639
15:16:00 dansmith melwitt: cool
15:16:23 melwitt thanks!
15:16:34 CeeMac i don't have cpu_mode set in nova.conf by the look of it
15:17:26 CeeMac and looking at the virsh dumpxml for an instance the cpu mode is 'custom' with a fallback model Skylake-Client-IBRS
15:18:50 sean-k-mooney CeeMac if you dont set it it default to None which in the libvirt driver is converted to host-model
15:18:55 melwitt lyarwood: easy fix for the cherry-pick hash check script if you get a moment https://review.opendev.org/756639
15:19:33 CeeMac sean-k-mooney: thanks, i think I recall reading something like that
15:19:45 CeeMac i can't seem to find where to set the trait in metadata
15:19:51 CeeMac could be veing blind though
15:20:52 sean-k-mooney openstack flavor set --property trait:HW_CPU_X86_VMX=retuired <flavor>
15:21:00 sean-k-mooney i think that is the correct command
15:21:04 sean-k-mooney stephenfin: ^
15:21:33 sean-k-mooney stephenfin: do you know if you have to quote the trait?
15:21:39 CeeMac ah yes, the old cli :)
15:21:49 sean-k-mooney ... retuired should be required
15:21:54 CeeMac got it
15:22:00 sean-k-mooney CeeMac: never trust my spelling
15:22:18 stephenfin sean-k-mooney: only if there are spaces or things the shell could confuse
15:22:29 stephenfin I usually quote to be safe
15:22:50 sean-k-mooney ya i often do but i was not sure if it would mess up the parsing of the =
15:24:22 CeeMac stephenfin: so "trait:.....required" <flavour>
15:25:14 stephenfin CeeMac: Yup. We list the supported versions here https://docs.openstack.org/nova/latest/configuration/extra-specs.html#trait
15:25:19 stephenfin *supported traits
15:25:41 CeeMac perfect, thanks
15:26:04 sean-k-mooney CeeMac: just be aware that we and them together
15:26:19 sean-k-mooney so if you use the amd trait and the intel trait then it wont find a host
15:26:45 CeeMac right, good to know :)
15:26:53 sean-k-mooney if you have a mix you should use a custom extra spec instead of a trait and use host aggreates
15:27:31 CeeMac i'm pretty sure at this point I don't want to do it at all, but i'll try it in my test env to be sure
15:27:41 CeeMac stephenfin: from that link it looks like I need trait{group}:HW_CPU_X86_VMX
15:27:55 CeeMac is the {group} specific to something?
15:28:02 sean-k-mooney there was a proposal to support "in" so you coudl do somehtin like trait:in(intel...,amd...)=required
15:28:13 sean-k-mooney CeeMac: you shoudl avoid groups
15:28:20 CeeMac i can do that
15:28:26 sean-k-mooney just leave out the group
15:28:32 CeeMac got it
15:28:36 sean-k-mooney we allow named or number grousp of triats
15:28:49 sean-k-mooney but they are hard to use and we may eventually remove it form the flaovr
15:29:15 CeeMac yeah, i'm not a big fan of "hard to use" so I'm ok with that :D
15:29:41 sean-k-mooney groups dont interact well when you have request form multipel sources
15:29:52 sean-k-mooney e.g. flavor, neutron and cyborg
15:30:42 sean-k-mooney there are some other gotchas so unless you know how its implmented you should avoid them
15:33:11 CeeMac i'll avoid them
15:33:46 CeeMac to be honest, i'd rather not allow nested virtualisation, but I'm being instructed to see if it is an option, so yeah
15:39:22 sean-k-mooney CeeMac: i always use nested vert and have done for years but its not for everyone
15:39:44 sean-k-mooney it has less sharp edges then it used too
15:40:29 CeeMac looks like OSA is set up to support it out of the box as well from a kernel module perspective
15:40:59 sean-k-mooney CeeMac:it was enabled by default in the upstream 4.19 kernel
15:41:02 CeeMac if it works without me having to make any major changes and reboot everything then I'm ok with that
15:41:05 CeeMac oh
15:41:13 CeeMac fair enough :)
15:41:43 CeeMac instance has deployed using a flavor with that trait
15:41:48 CeeMac so looking fine so far
15:42:46 CeeMac its more of a legacy issue with my user base than a technical objection
15:43:26 CeeMac a lot of them still don't understand that IaaS means they're on their own unless they want to pay for support services
15:45:33 sean-k-mooney ya
15:46:06 sean-k-mooney the old issue wit nested virt was you could not migrate the l1 vm if the l1 guest was running level 2 guests
15:46:30 CeeMac ah, yeah i could see that being a problem
15:47:14 sean-k-mooney its been fix in 4.19
15:47:54 sean-k-mooney that was on of the technical reason operator did not often use it in the past
15:48:11 sean-k-mooney since it limited how they could use live migration
15:50:01 CeeMac makes sense
15:58:58 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Remove support for '[libvirt] use_usb_tablet' https://review.opendev.org/756550
15:58:58 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Only add a USB controller if it's necessary https://review.opendev.org/756549
15:58:59 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Add support for virtio-based input devices https://review.opendev.org/756552
15:58:59 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Rationalize attachment of USB tablet https://review.opendev.org/756551
15:59:59 sean-k-mooney stephenfin: did you add None to the pointer device. the original requirement diring this is to have no input device attached to the guest
16:00:11 sean-k-mooney so using virtio wont actully help in the realtime case
16:00:27 sean-k-mooney i assume that is just a nice to have while your here
16:03:11 admin0 hi guys .. can the default domain be null or "" blank ?
16:03:15 admin0 insetad of .novalocal ?
16:03:33 sean-k-mooney dns domain
16:03:44 admin0 can it be null ?
16:03:56 admin0 or some value has to be there
16:04:14 sean-k-mooney its set by https://docs.openstack.org/nova/latest/configuration/config.html#api.dhcp_domain
16:04:51 admin0 it does not say if it can be left null
16:05:00 sean-k-mooney i dont think it can be
16:05:48 sean-k-mooney although we have exampel of it being empty
16:06:27 sean-k-mooney we use it like this
16:06:28 sean-k-mooney https://github.com/openstack/nova/blob/58636d34ec68d44952b22947bac01d4b88ae103a/nova/api/metadata/base.py#L533-L536
16:06:56 sean-k-mooney you would have to set it to ""
16:06:58 sean-k-mooney https://github.com/openstack/nova/blob/85c35b4d9d6386866ddf197e4a751f729b726188/nova/conf/api.py#L218-L222
16:07:04 sean-k-mooney but i dont know if that wil work or not
16:07:14 sean-k-mooney default might still replace ""
16:07:18 sean-k-mooney but it would be easy to test
16:07:34 sean-k-mooney just set [api]/dhcp_domain=""
16:12:42 melwitt it can be blank. we had it set that way at yahoo
16:35:07 stephenfin sean-k-mooney: No, it wasn't helpful. You can't disable the built-in PS2 devices
16:35:23 stephenfin However
16:35:38 sean-k-mooney you cant by generating model=none in the xml
16:36:00 stephenfin I don't think that's supported. The only RFE I found for it was DEFERRED
16:36:40 sean-k-mooney well the perfromacne team teh filed teh orignal downstream issue had said they disabeld it manually by editing the xml
16:37:02 stephenfin they were talking about USB devices
16:37:15 sean-k-mooney yes
16:37:32 stephenfin you won't get a USB keyboard or pointer device on x86 unless you request them via image metadata properties or nova.conf

Earlier   Later