Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-11-19
15:12:22 stephenfin so, revert the change to nova, merge the Python 3-ification patch to DevStack, wait a month and re-revert that change to nova
15:12:39 dansmith stephenfin: just the half-revert
15:12:40 sean-k-mooney we can trivially but we need to wait as a lib project
15:12:42 gmann revert only setup.cfg and req right ?
15:12:51 gmann yeah what dansmith mentioned
15:13:04 dansmith stephenfin: revert the min, default devstack to py3, later un-revert the minimum
15:13:12 gmann +1
15:13:31 stephenfin what about your concerns that we could accidentally break Python2 support in that time?
15:13:36 stephenfin don't they still hold?
15:13:39 dansmith stephenfin: those aren't my concerns
15:13:50 dansmith stephenfin: my concerns are that we break the _default_ config in devstack
15:13:58 dansmith if the default changes, I'm good
15:13:59 stephenfin ah
15:14:39 corvus i agree with dansmith (that's my concern too). but it's unclear to me whether we can immediately change the default. is that something we can do today?
15:14:52 stephenfin corvus, tosky: have you been following along or do you need a recap?
15:14:54 stephenfin whoops
15:14:55 dansmith sounds like gmann is willing
15:14:58 gmann stephenfin: you are updating devstack patch? or should i do
15:15:15 gmann gate will be fixed by nova revert
15:15:20 stephenfin gmann: I can...if you tell me what needs to be done?
15:15:24 dansmith corvus: we're going to do the half revert now as well, which I think will eliminate the actual devstack breakage immediately anyway
15:15:26 gmann so default change happen today or tomorrow should be ok
15:15:28 sean-k-mooney corvus: i think we can change the default today
15:15:31 tosky stephenfin: a bit swamped with meetings, I think I may have got the gist of it though
15:15:37 gmann stephenfin: sure, i will comment on patch
15:15:41 tosky like: some reverting will land for now
15:15:50 corvus there was a bunch of devstack centos7 stuff earlier that i didn't understand, but if people who do understand agree that's all doable, then cool. :)
15:15:51 tosky and some other work is needed
15:15:56 sean-k-mooney if it cause gate failure then project can set it back to false in each job that broke
15:16:00 tosky which is a bit generic summary but I'm sure it fits :)
15:16:12 stephenfin tosky: got it in one
15:16:20 dansmith stephenfin: I can summarize on the half-revert patch and +2, you can come back and +W when you're done with the devstack thing okay?
15:16:27 stephenfin ack
15:16:41 gmann corvus: i pushed patch to add centos8 job so i think we can drop centos7 completely
15:17:02 sean-k-mooney corvus: centos ships with python 3.4 so it need to have py 3.6 installed form eple or we need to swap master jobs to centos 8
15:17:26 corvus and centos8 is ready?
15:17:28 tosky gmann: see on #openstack-qa: some work is needed on centos 8, frickler was working on it
15:17:28 stephenfin gmann shows up and fixes everything for us. This is excellent \o/
15:17:31 sean-k-mooney we just need to fix any gaps in devstack on centos 8
15:17:32 gmann my bad and sorry for not considering the setup.cfg min py version things during composing the plan/goal
15:17:45 tosky gmann: and the requirements too :)
15:17:51 gmann yeah :(
15:18:11 stephenfin gmann: Not your fault. I never thought about the DevStack default thing when putting that together
15:18:40 mriedem stephenfin: your patch depended on the devstack default switch after i mentioned it, and then you dropped the depends-on
15:18:54 mriedem likely because we could just set USE_PYTHON3=True in our jobs
15:19:05 stephenfin yup, exactly because of that :(
15:19:40 stephenfin I thought the depends on was an optimization to avoid us having to set that stuff manually
15:19:48 stephenfin TIL
15:19:56 tosky gmann: no problem, really; it happens, let's improve the process and continue
15:20:14 mriedem slaweq: do you have a bug for this yet so i can track it in e-r
15:20:15 mriedem ?
15:20:54 slaweq mriedem: no, I didn't open any bug for this
15:21:10 slaweq I just checked here that You're already talking about it so I didn't report a bug
15:21:14 corvus i think the thing to watch out for in future reviews is: when setting a devstack option, is it something that is intrinsic to the specific job (okay in nova) or is it something that is recommended/required for any devstack users (should be in devstack). if all the nova devstack jobs have an option, it's almost certainly the latter.
15:23:50 stephenfin makes sense
15:29:17 dansmith stephenfin: https://review.opendev.org/#/c/695007/3
15:29:23 dansmith efried: mriedem ^
15:31:37 stephenfin dansmith: LGTM
15:32:04 efried dansmith: so you've changed your mind on readding a minimal py2 job?
15:32:23 dansmith efried: no, you must not have followed all of that, and.. did you read my wall of text?
15:32:27 efried I did
15:32:52 dansmith efried: all I care about is that we're testing the devstack default.. if the default is changing, I'm cool with sticking with py3 only and letting py2 issues creep in
15:32:58 sean-k-mooney efried: im guessing your askign since this patch does not add the minimal job
15:33:03 dansmith efried: if the default is not changing, I think we have to test and validate the default
15:33:36 efried what sean-k-mooney said. since this patch doesn't depend on the "switch the default to py3" patch, and there's no telling how long it will be until that actually merges, we have The Hole.
15:34:00 mriedem you have a gentlemans agreement for about the next 24 hours
15:34:00 dansmith efried: gmann said they'd merge it today
15:34:11 efried okaaay
15:34:11 mriedem after that courtney love shows pu
15:34:12 mriedem *up
15:34:14 dansmith we could block on that, but we're blocking a bunch of other people
15:34:35 efried yah, seems like it would have been easy to add one line to .zuul.yaml in that patch, but mebbe not.
15:34:44 dansmith gmann: don't f*ck me over, okay?
15:34:45 efried anyway, it's on the way
15:34:47 mriedem for me it's not about ease
15:34:48 dansmith efried: well, I'm fine with that if you want
15:34:50 sean-k-mooney efried: nova-dsvm-multinode-base is still python 2 i think
15:35:01 mriedem sean-k-mooney: that's not a ojb
15:35:02 mriedem *job
15:35:04 mriedem that's a base
15:35:12 sean-k-mooney so nova-live-migration will by python 2
15:35:16 sean-k-mooney ya i know
15:35:16 mriedem it's not
15:35:19 efried nope
15:35:31 stephenfin sean-k-mooney: it can't be. if it was, it wouldn't work for us :)
15:35:33 mriedem hi https://review.opendev.org/#/c/694787/
15:36:07 sean-k-mooney im looking at https://github.com/openstack/nova/blob/master/.zuul.yaml and i dont see python 3 enabled
15:36:08 efried because https://opendev.org/openstack/nova/src/branch/master/playbooks/legacy/nova-live-migration/run.yaml#L50 rite?
15:36:16 mriedem it's in the legacy playbook
15:36:17 mriedem yes that
15:36:37 mriedem cue sean-k-mooney saying we should really move those jobs to zuulv3
15:36:41 stephenfin mriedem: on that, you need to keep the sphinx on py2 line in doc/requirements.txt
15:36:50 stephenfin i tried removing it and got the same thing you're seeing
15:36:53 stephenfin the tool is dumb
15:37:17 sean-k-mooney ah ok
15:37:22 gmann ? anything changed ?
15:37:44 dansmith gmann: no, we're just taking bets on whether or not you're really going to merge the devstack change today
15:37:52 openstackgerrit Matt Riedemann proposed openstack/nova master: FUP to Ie1a0cbd82a617dbcc15729647218ac3e9cd0e5a9 https://review.opendev.org/694787
15:38:17 dansmith gmann: efried and I will have to meet on the playground after school if not, and he'll kick my ass
15:38:28 gmann ok :)

Earlier   Later