Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-01-22
15:41:11 jaypipes sean-k-mooney, stephenfin: ooooh! on that wikipedia page I've officially found my new favorite Irish word derived from old English: "Eejit".
15:42:51 sean-k-mooney eejit is a great word
15:45:40 sean-k-mooney usually used to refer a harmless idoit or bafoon or equavlent to the irish Amadán pernounced "om a dawn"
15:45:50 jaypipes Also, I have gained some new knowledge: Irish lacks words that directly translate as "yes" or "no"
15:46:38 jangutter jaypipes: That's probably why Nancy Reagan's anti-drug campaign failed there.
15:47:17 mriedem jaypipes: did you sort out that rebuild with double allocation NoValidHost thing? like i said, i'm pretty sure that's been fixed but maybe you don't have the backport
15:49:19 jaypipes mriedem: yup, between yourself and melwitt, found the right patch and am applying it to our ocata downstream this morning. thank you!
15:49:25 jaypipes jangutter: lol!
15:52:13 jangutter sean-k-mooney: my favourite word there is "banjaxed", it's just under my all-time favourite word, "defenestration".
15:52:34 stephenfin `<jaypipes> sean-k-mooney: that's interesting, considering there are not distinct words for "three" and "tree"` :D
15:53:14 jangutter sean-k-mooney: The kuryr-kubernetes-tempest-daemon-octavia job is banjaxed, it should be defenestrated.
15:53:41 openstackgerrit Matt Riedemann proposed openstack/nova master: Convert port to str when validate console port https://review.openstack.org/626853
15:56:21 sean-k-mooney jangutter: stephenfin jangutter im adding some examples of the k8s jobs issue by the way.
15:56:52 jangutter sean-k-mooney: yeah, I've seen a number of them that really should not have been triggered by the os-vif changes.
15:59:36 jaypipes stephenfin: I see at least one person appreciated my joke :)
15:59:50 sean-k-mooney jaypipes: i did too :)
16:01:46 jaypipes sean-k-mooney: :)
16:01:54 jaypipes sean-k-mooney: well, that makes tree of us, then.
16:05:53 sean-k-mooney updated with 5 differnet issue that have required a recheck in the last week https://review.openstack.org/#/c/632484/3
16:09:50 sean-k-mooney huh the k8s schduler service failed to start on the patch to make it non voting... http://logs.openstack.org/84/632484/3/check/kuryr-kubernetes-tempest-daemon-octavia/4bcca98/job-output.txt.gz#_2019-01-22_15_55_22_393153
16:11:27 mriedem zzzeek: this should unblock the sqla-migrate release https://review.openstack.org/#/c/632470/
16:12:26 zzzeek +2ed
16:15:04 sean-k-mooney ralonsoh: jangutter by the way if the current recheck of https://review.openstack.org/#/c/629025/ and https://review.openstack.org/#/c/632077/2 fail can you leave them be until we decide what to do with the k8s job
16:15:34 jangutter sean-k-mooney: yah, not touching 'em.
16:16:09 sean-k-mooney ralonsoh: jangutter im watching them on the zuul status page si ill eiterh rebase or recheck depending on what we decide
16:21:28 jangutter sean-k-mooney: etcd is also not happy: http://logs.openstack.org/84/632484/3/check/kuryr-kubernetes-tempest-daemon-octavia/4bcca98/controller/logs/screen-etcd.txt.gz
16:22:04 sean-k-mooney jangutter: ya ectd proably is why the scheduler did not start
16:23:18 sean-k-mooney i check the logs everytime the job failes but i always seam to root cause it to a setup issue or something unrelated to the patch.
16:23:57 sean-k-mooney dont get me wrong we could break kuryr kubernetese with some change to the ovo but not unit test chagnes
16:24:50 sean-k-mooney i have a todo to add more non-voting jobs to os-vif which i hope to make voting after stien
16:33:57 openstackgerrit Kashyap Chamarthy proposed openstack/nova master: [WIP] libvirt: Bump MIN_{LIBVIRT,QEMU}_VERSION for "Stein" https://review.openstack.org/632507
16:35:46 kashyap (If anyone wonders why "WIP" (Work In Progress) and why not Workflow -1 — WIP spares people from clicking on the URL when they see it on IRC.)
16:36:48 kashyap (And putting the Workflow-1 will make the change not appear for those who use dashboards. So using both is good.)
16:36:54 stephenfin mriedem: That has to be a bug with PyPI
16:36:58 stephenfin mriedem: Wait, no. Commented
16:39:06 stephenfin mriedem: Is moving from sqlalchemy-migrate to alembic a thing we'd want to do at some point in the future. Have discussed roughly how to do it with zzzeek and can attempt it at some point in the next few months
16:39:10 stephenfin *?
16:39:26 mriedem stephenfin: not in nova probably not
16:39:29 mriedem thanks for the docs thing
16:39:39 mriedem back in a few hours (funeral)
16:39:45 stephenfin o/
16:41:09 sean-k-mooney stephenfin: maybe add it as a train ptg topic. it has been mentioned before i think but i dont think it was ever seen as a high priority in the past
16:45:38 stephenfin sean-k-mooney: That's why I asked. It came up due to bnemec noting that it was effectively dead and zzzeek saying that alembic was a far more capable replacement
16:46:05 stephenfin sean-k-mooney: Sounds similar to the mox-mock thing though, and I don't want to get involved in another one of those /o\
16:47:23 kashyap Cleaning up these unit test excrements after version libvirt / QEMU bump is just a bottomless pit...
16:47:26 kashyap "Endless fun"
16:48:29 zzzeek mriedem_away: you wouldn't accept the effort by a volunteer even ?
16:48:42 sean-k-mooney stephenfin: we discussed that neutron uses alembic i think and should nova also move to it but it was a non trivial amount of work and madde ffu even harder
16:49:15 zzzeek sean-k-mooney: the placement service was broken off from nova and uses alembic. this was already a mini-migration of former parts of nova from migrate to alembic
16:49:19 sean-k-mooney so i think we jsut said maybe in the futre but since its not causeing us issue not done "fix" it
16:50:08 sean-k-mooney zzzeek: i remember cdent and i spoke about that brefily in denver actully
16:50:38 zzzeek sean-k-mooney: nova cant be on sqlalhcemy-migrate forever because I can't keep supporting migrate forever from my side
16:50:49 sean-k-mooney zzzeek: i think in stien it just was not high enough on our priority list
16:51:29 sean-k-mooney zzzeek: sure and as a maintainer if that is deperacted we shoudl come up with a transition plan so you dont have too but it would need a spec and need to be a train thing i think
16:51:33 zzzeek sean-k-mooney: im all for it being bottom priority but that made me nervous just now when mriedem_away said it was a "never" thing
16:51:58 zzzeek sean-k-mooney: if you look at what placement did, it's that
16:52:51 sean-k-mooney nova has a lot more legacy stuff to deal with.
16:54:02 sean-k-mooney zzzeek: as i said its something we should talk about for train.
16:54:31 sean-k-mooney zzzeek: is sqlalcamy-migrate offially deprecated/unsupproted
16:54:42 zzzeek sean-k-mooney: if nova got off it, it sure would be
16:54:57 zzzeek sean-k-mooney: some other projects too. keystone cinder not sure if they are still on it
16:55:16 zzzeek sean-k-mooney: it's "supported" because we have no choice
16:55:32 zzzeek sean-k-mooney: also there are no "legacy" cases that alembic doesn't support from a DB perspective
16:55:35 sean-k-mooney right which give nova/openstack 2 options. 1 they fork/take over maintance of it 2 the move to something else.
16:56:02 zzzeek sean-k-mooney: openstack already owns sqlalchemy-migrate
16:56:20 sean-k-mooney oh ok
16:57:11 zzzeek sean-k-mooney: I only know what sqlalchemy-migrate even looks like becuase I joined red hat and got assigned to it :)
16:57:25 sean-k-mooney ah i see.
16:57:27 zzzeek the deep irony that I end up maintaining a project that I replaced w/ somehting else
16:58:24 smcginnis zzzeek: Cinder is still using it. What should we be switching to?
16:58:42 zzzeek smcginnis: alembic migrations have superseded sqlalchemy-migrate for many years
16:59:48 smcginnis zzzeek: Ah. I had raised that as a question ~1 year ago and was shot down as too complex to switch and not really needed.
16:59:52 stephenfin kashyap: Oh, that's why it's WIP. Might stick that in the commit message on the next revision, if you're doing this in a piecemeal fashion
16:59:53 sean-k-mooney zzzeek: its a little technology specific but perhaps it coudl be raise as a cross project goal
17:00:07 zzzeek smcginnis: yeah it's like I am being too nice by not just breaking compatibility with it or something
17:00:10 smcginnis But I can raise it again now with this. We have a midcycle coming up that will be a good venue to discuss it.
17:00:39 zzzeek smcginnis: im looking for reason to go back to denver in may so if you think I'd be useful send me a ping
17:00:54 sean-k-mooney smcginnis: i think that was the same sentime it was discussed with in nova in the past too
17:01:00 smcginnis sean-k-mooney: ++
17:01:17 zzzeek sean-k-mooney: technically, I was supposed to be getting everyone on alembic years ago, that was part of my job duties. but I get the big wall of "we dont care"
17:01:29 smcginnis zzzeek: Since there are a few projects that have not switched, it might be good to have a Forum or PTG session to go over how to migrate.
17:02:01 kashyap stephenfin: Will add that note. I have a couple of other patches related to it. Meanwhile, I'm trying to figure out the first unit failure of "testtools.matchers._impl.MismatchError" in the test_get_guest_config.
17:02:14 zzzeek smcginnis: yeah I guess you are right
17:03:05 smcginnis zzzeek: We could also try to make it a cycle goal, so having a session to kick that off could really help there.
17:03:28 sean-k-mooney zzzeek: from a redhat point of view we have an lts every 3 releases so makeing the switch in Train/osp16 would be good form a support perspective on your part
17:03:52 sean-k-mooney zzzeek: by the way i know you know that :)
17:04:01 zzzeek smcginnis sean-k-mooney OK....but if I just do a session and then, everyones like OK mike go do it, thats not really going to work for me :)
17:04:25 smcginnis Oh come on, we would love it if you did all the work. :D
17:04:35 zzzeek sean-k-mooney: the problem is this is strict technical debt stuff nobody cares
17:04:52 zzzeek smcginnis: even if I *do* all the work, my reviews get ignored becuase they make everyone nervous
17:05:03 zzzeek smcginnis: so it's not even doing the work
17:05:07 zzzeek its the huge wall of indifference
17:05:48 smcginnis It's a really big wall.
17:06:19 zzzeek smcginnis: this is why, somehow, when they did placement, enough people on "placement" cared to do the switch, was refreshing to see
17:06:44 zzzeek but i think it's important that people who are core to the projet are involved in the work. nobody trusts me
17:07:43 smcginnis Good to start socializing the idea now then. And maybe another good reason to try for a session in Denver to make sure folks are aware and understand that this is technical debt that really does need to be addressed.
17:08:15 sean-k-mooney zzzeek: if you propose patches stephenfin will likely review them and if you talke about why the change is needed and point ot plamcent and other that have done it already people will trust your work
17:09:00 stephenfin ^ this is true
17:09:24 stephenfin Will need a bit of hand holding to avoid "my reviews get ignored becuase they make everyone nervous" though
17:09:48 sean-k-mooney zzzeek: i can try and review it too but im not a core and i usually avoid the db layer

Earlier   Later