Earlier  
Posted Nick Remark
#openstack-nova - 2023-02-28
16:11:10 elodilles with a -1 on it
16:11:33 bauzas at least we need to merge first https://review.opendev.org/c/openstack/nova/+/874932 (min servion version), https://review.opendev.org/c/openstack/nova/+/873584 (logging revert) and https://review.opendev.org/c/openstack/nova/+/875380 (reno prelude)
16:11:41 elodilles (and thanks in advance o:))
16:11:58 bauzas elodilles: indeed, doing it now
16:12:39 bauzas so
16:13:20 bauzas about https://review.opendev.org/c/openstack/nova/+/874932 I think we are OK with how to use rolling upgrades with SLURP and non-SLURP releases
16:14:12 bauzas sean-k-mooney had a concern with https://review.opendev.org/c/openstack/nova/+/875380 but we'll discuss it tomorrow
16:14:45 bauzas anything people want to discuss with RC1 ?
16:15:22 gibi bauzas: will you add a reno for https://review.opendev.org/c/openstack/nova/+/874932
16:15:47 gibi bauzas: it now allows Antelope to support Yoga computes
16:15:53 bauzas gibi: I provided a new phrase in the prelude
16:15:54 gibi which is N-2
16:16:18 bauzas gibi: but like you said, we could also modify the rolling-upgrade doc
16:16:48 gibi yeah, my point is we either do a proper documented N-2 support, or we keep the N-1 check for now
16:16:51 bauzas prelude patch is https://review.opendev.org/c/openstack/nova/+/875380
16:17:03 gibi but right now the code patch allows N-2 while our doc says otherwises
16:17:22 bauzas gibi: okay, then I'll add a new change before we create the RC1
16:17:34 gibi in the current form I cannot +2 https://review.opendev.org/c/openstack/nova/+/874932
16:17:49 gibi as it is inconsistent with our doc
16:18:15 gibi but as I said I'm OK to move the doc to match with the code or move the code to match with the doc
16:18:16 bauzas gibi: so you want me to modify https://docs.openstack.org/nova/latest/admin/upgrades.html#rolling-upgrade-process ?
16:18:51 gibi bauzas: I would like to see a consistent documentation about Antelope supports regarding old compute service versions
16:19:13 gibi if we go with Antelop supporting Yoga computes then yes https://docs.openstack.org/nova/latest/admin/upgrades.html#rolling-upgrade-process needs a change
16:19:45 gibi (or we just fall back to only support N-1 and modify https://review.opendev.org/c/openstack/nova/+/874932 to only support Zed computes in Antelope)
16:19:56 bauzas cool, then I'll just add a doc modification for it in https://docs.openstack.org/nova/latest/admin/upgrades.html#rolling-upgrade-process
16:20:00 bauzas shit
16:20:11 bauzas add a doc modif in https://review.opendev.org/c/openstack/nova/+/874932
16:20:23 gibi OK
16:20:55 bauzas do people prefer here to only support Zed instead of Yoga ? https://review.opendev.org/c/openstack/nova/+/874932
16:21:00 bauzas dansmith: sean-k-mooney: ^
16:21:25 dansmith for bobcat or antelope?
16:21:51 bauzas MHO is that I'd prefer to modify the rolling upgrade document and say yes, we can support Yoga computes for Antelope services
16:21:55 dansmith as I think I've mentioned, for antelope I think it would be nice to have a best-effort trial run of N-2 support, so antelope would support yoga
16:22:04 dansmith yeah that ^
16:22:14 sean-k-mooney ya i think yoga makes sense
16:22:33 bauzas dansmith: yeah, gibi asked either to modify the doc or not modify it but only support Zed computes with Antelope conductors
16:22:44 bauzas I'd prefer the former
16:22:48 dansmith yeah
16:22:49 sean-k-mooney with that said for bobcat technially it should be antelope but i assume dansmith would prefer zed
16:23:09 dansmith sean-k-mooney: I'd prefer we just try to do N-2 always unless we come up with a blocking reason we can't yeah
16:23:22 bauzas that's another change https://review.opendev.org/c/openstack/nova/+/875621/ that needs to be merged *after* RC1
16:23:23 dansmith I think we'll be able to do it
16:23:35 sean-k-mooney right whihc is not strictly requried by the governance resolution but im ok to try that until it breaks
16:23:48 bauzas but looks like dansmith prefers to support Zed computes with Bobcat so meh
16:23:48 gibi then modify the upgrade doc to state that in Antelope we support N-2 (Zed) computes as best effort
16:23:53 dansmith sean-k-mooney: if nothing else, it'll highlight the thing(s) that prevent us from doing it, which are good for planning
16:24:21 dansmith gibi: best effort is the right terminology yeah
16:24:27 bauzas ok, so let's try to make the grenade-skip-level job voting
16:24:48 dansmith s/make/keep/ right?
16:24:56 bauzas it's n-v as I speak
16:25:05 sean-k-mooney voting based on 22.04 wiht zed ->bobcat
16:25:06 bauzas and only in check
16:25:08 dansmith oh, I thought I looked
16:25:20 bauzas only in check pipeline
16:25:21 dansmith okay, I see.. yeah "make" then
16:26:01 bauzas sean-k-mooney: cool then I can modify https://review.opendev.org/c/openstack/nova/+/875621/ to make grenade-skip-level *voting*
16:26:28 bauzas (and adding it to the gate pipeline)
16:27:06 bauzas that one would be merged *AFTER* 2023.1 RC1 to be clear, so it would be a 2023.2 change
16:27:17 bauzas agreed ?
16:27:43 sean-k-mooney yes
16:27:58 bauzas ok, and about your point with focal, sure
16:28:21 bauzas we need to test the N-2 rolling upgrade with ubuntu 22.04 only, right?
16:28:30 sean-k-mooney yes
16:28:39 bauzas cool
16:28:41 sean-k-mooney because i want us to bump our min libvirt/qemu
16:28:45 bauzas looks like we have consensus
16:28:51 dansmith I think we can only switch that once we convert it to start from zed, which it doesn't currently do
16:29:02 sean-k-mooney and i htink the ones we chose last were released in 20.10
16:29:24 sean-k-mooney yes it will need to wait for changing to zed
16:29:27 bauzas what's the TC supported release for Yoga ?
16:29:29 sean-k-mooney which might need a grenade change
16:29:41 dansmith sean-k-mooney: it will yeah
16:29:53 sean-k-mooney https://github.com/openstack/governance/blob/master/reference/runtimes/yoga.rst
16:29:57 sean-k-mooney 20.04
16:30:19 sean-k-mooney its technically the same for zed 20.04
16:30:28 sean-k-mooney but canoncial released zed on 22.04
16:30:50 bauzas that's what I was looking for
16:30:51 dansmith I need to talk to gmann first but I will propose the skip-level job change
16:30:59 bauzas so
16:31:19 sean-k-mooney devstack works fin with zed on 22.04 the last time i tried it so i expect it to work
16:31:21 bauzas are we saying that we would use focal or jammy for the N-2>N skip-level job ?
16:31:35 sean-k-mooney focal for now jammy when we change to zed
16:31:40 dansmith jammy once we upgrade to zed->bobcat
16:31:43 bauzas cool
16:31:54 bauzas I agree then
16:32:24 bauzas #action bauzas to make our grenade-skip-level job voting after antelope rc1 on focal
16:32:39 bauzas I guess we're done with that topic
16:32:46 bauzas last point tho
16:32:54 bauzas #info who's interested in running for being a release liaison ?
16:32:54 bauzas #info who's interested in running for being a release liaison ?
16:32:59 bauzas we discussed it last week
16:33:11 bauzas as a reminder, https://docs.openstack.org/project-team-guide/release-management.html#release-liaisons
16:33:48 bauzas tl;dr: look at Gerrit to see whether the releases team created a patch for reviews
16:33:58 bauzas and say yay or nay on that patch
16:34:35 bauzas that job is basically backed by a French person that knows it for a while
16:34:54 bauzas but I'd prefer not to be alone in order to avoid elodilles to hassle me with pings
16:34:56 bauzas :p
16:35:07 elodilles note: release liaisons are added as reviewers automatically (usually) to related release patches
16:35:36 bauzas on a side note, I have a non-urgent patch for cleaning up the PTL guide https://review.opendev.org/c/openstack/nova/+/875730
16:35:40 sean-k-mooney i have done it now for 2-3 releases i can try and find time when pinged to look but if other are intereested let us know

Earlier   Later