Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-11-05
19:26:30 mriedem that works for me.
19:26:31 melwitt because actually making it correct for osc is not straightforward
19:26:37 mriedem btw, i thought rhosp had it's own docs and didn't use upstream?
19:26:43 melwitt and would be better off as a separate change
19:27:36 amodi mriedem: melwitt sure, updting the patch to use nova boot
19:28:08 amodi yes, we dont use upstream, except when we do :)
19:28:18 mriedem btw, this was the breaking change in the manuals repo https://review.opendev.org/#/c/404623/
19:29:14 melwitt yeah, that was part of the big sweep that did some find/replaces. other docs of ours became wrong in that sweep, can't remember which atm
19:29:25 melwitt oh, the quota stuf
19:29:30 melwitt stuff
19:30:39 mriedem yup the quota one was annoying
19:30:49 mriedem i left an updated comment on amodi's change hopefully clearing it up
19:31:05 mriedem amodi: btw, you should create a doc bug for this since we'll backport the fix
19:31:41 mriedem and now i'm off to vote on raising my taxes for schools because i love the children
19:32:09 amodi mriedem: ack, will do, and im just updating this doc
19:32:32 amodi or if u want me to update all the other ones in the referenced changed, lmk
19:33:06 mriedem amodi: just that doc. i didn't look through https://review.opendev.org/#/c/404623/ to see if it broke other stuf
19:33:07 mriedem *stuff
19:33:38 melwitt it broke the boot from ISO image too (still looking)
19:33:43 amodi https://review.opendev.org/#/c/404623/8/doc/user-guide/source/cli-nova-launch-instance-using-ISO-image.rst this has the --block-device used with osc command
19:34:26 melwitt the other stuff is probably ok
19:34:31 mriedem ah yeah that is wrong, osc before 4.0 didn't have a way to boot with an image and create a blank volume
19:34:37 mriedem yeah the rest looked ok
19:35:15 amodi cool, will update thanks!
19:35:22 mriedem actually i'm not even sure latest osc allows creating a server and telling nova to create and attach a blank volume
19:35:45 mriedem add it to the gap pile!
19:35:51 mriedem https://etherpad.openstack.org/p/compute-api-microversion-gap-in-osc
19:36:40 mriedem artom: you broke nothing https://review.opendev.org/#/c/693045/
19:36:51 openstackgerrit Boris Bobrov proposed openstack/nova master: Create a controller for qga when SEV is used https://review.opendev.org/693072
19:36:51 openstackgerrit Boris Bobrov proposed openstack/nova master: Also enable iommu for virtio controllers and video in libvirt https://review.opendev.org/684825
19:37:01 melwitt hoo that is a lot of gaps
19:38:16 mriedem (1) i thought you were artom (2) saying that is a lot of gas
19:40:09 melwitt osc might cause gas
19:40:19 artom mriedem_afk, I know, I suck
19:40:33 artom Or... blow? Which way is the gas supposed to go?
19:40:55 dansmith suck squeeze bang blow
19:41:08 artom That's the one
19:41:15 artom Applies to ICEs and other things
19:41:30 dansmith why do you think mriedem_afk is afk right now?
19:41:32 dansmith HEY-O
19:41:59 dansmith oh, voting, nevermind
19:42:36 artom In some ways it's the same thing
19:42:51 dansmith heh
19:47:04 artom I also like how he keeps -1'ing the _DNM_ change :)
19:58:55 openstackgerrit Artom Lifshitz proposed openstack/nova master: DNM: break live migration to trigger bug 1788014 https://review.opendev.org/693045
19:58:56 openstack bug 1788014 in OpenStack Compute (nova) "when live migration fails due to a internal error rollback is not handeled correctly." [Medium,In progress] https://launchpad.net/bugs/1788014 - Assigned to Matt Riedemann (mriedem)
20:07:36 sean-k-mooney artom: its possibel the breakage has something to with vhost-user other then the mtu or libvirt could have fixed the issue and now always use the 64bit encoding instead
20:09:30 artom sean-k-mooney, I added an extra debug log, we'll see
20:09:39 artom If the latter, we can try and find another way to trigger this
20:09:46 sean-k-mooney it been over a year since i actully hit that error and it was on cent7.5 so the libvirt we are using in the gate in the ubuntu vms is alot different
20:10:47 efried sean-k-mooney: what do we do about tempest tests that need a certain minimum of e.g. libvirt or qemu?
20:11:06 efried Do all the CI nodes have the same (latest) version?
20:11:21 efried hm, I guess they probably do if they all use the same image...
20:11:22 sean-k-mooney it depends
20:11:45 sean-k-mooney in some cases we have feature flags in the tempest config
20:11:56 sean-k-mooney but in general tempest does not have test that care
20:12:29 sean-k-mooney in the nova code we have checks if feature need a newer version then our minium
20:12:35 efried I'm going to be getting into vTPM with encryption this cycle, which requires a very recent libvirt/qemu. I don't know yet whether I'll be adding CI for it, but was curious how that would happen if I did.
20:12:59 efried yeah, but that'll just cause the build to fail, right?
20:13:16 sean-k-mooney am its a pain but we used to use fedora with the virt preview repo for that
20:13:47 sean-k-mooney if that is not an option we also have a devstack plugin you could use
20:13:58 sean-k-mooney it will compile libvirt and qemu from source
20:14:02 efried eek
20:14:36 sean-k-mooney personally i found that beeter then random fedora virt preview package
20:14:52 efried I don't think I need *unreleased* versions. Just pretty recent ones.
20:14:53 sean-k-mooney efried: do you know what version you need
20:15:01 efried yeah, one sec...
20:16:10 efried sean-k-mooney: libvirt 5.6.0. qemu at least 1.11, but possibly higher, not sure.
20:16:14 efried sorry, 2.11
20:16:43 sean-k-mooney ok fedroa 20 might have that 5.6 is pretty new i think
20:16:46 sean-k-mooney https://opendev.org/x/devstack-plugin-libvirt-qemu
20:16:54 sean-k-mooney i can try and build those with that too
20:17:01 sean-k-mooney i havent really been maintaining that
20:17:12 sean-k-mooney but its what we used to use at intel to test prerelease stuff
20:18:17 mriedem artom: -1ing your DNM changes is how i stay up on stackalytics
20:21:53 mriedem the bottom 2 trivial changes in the cross-cell resize series are begging for another core https://review.opendev.org/#/c/692856/
20:22:03 mriedem they are, not me
20:22:56 sean-k-mooney efried: ill test it but i think you just need to add http://paste.openstack.org/show/785823/ for the devstack plugin
20:26:38 openstackgerrit Archit Modi proposed openstack/nova master: Revert "openstack server create" to "nova boot" in nova docs https://review.opendev.org/693065
20:30:21 efried whoah, I appear to have 4.0.0 on my bionic
20:30:32 efried is the apt package level the number I want to be looking at?
20:30:50 sean-k-mooney 4.0 of?
20:31:02 sean-k-mooney libvirt you check with libvirtd --version
20:31:23 efried yeah, 4.0.0
20:31:51 sean-k-mooney ya and ubuntu 18.04 has 2.11 libvirt
20:32:00 sean-k-mooney *qemu
20:32:29 sean-k-mooney so the qemu in the gate should be fine
20:32:35 efried confirmed 2.11
20:33:04 efried Is there a way *other* than building from source (which I have chronically shit luck with, no matter what I'm trying to build) to get a higher libvirt on my bionic?
20:33:39 sean-k-mooney i dont think there is a ppa but it migth be in the ubuntu cloud archive
20:33:43 sean-k-mooney let me check
20:34:39 sean-k-mooney ya they have 5.4
20:34:47 sean-k-mooney http://ubuntu-cloud.archive.canonical.com/ubuntu/pool/main/libv/libvirt/
20:35:04 sean-k-mooney you need 5.6 however right
20:35:05 efried okay, so if I need 5.6.0 I need to build
20:35:06 efried yeah.
20:35:15 efried wah wah wahhhhh
20:36:46 sean-k-mooney even the libvirt ppa is 5.4 https://launchpad.net/ubuntu/+source/libvirt
20:37:20 sean-k-mooney ill let you know in a few minute i the plug still work.
20:37:57 efried sean-k-mooney: I gotta do this locally first anyway

Earlier   Later