Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-12-03
15:01:02 jangutter is there a separate mailing list for os-vif (or is it mostly nova + neutron)...
15:01:42 sean-k-mooney i will likely propsoe a patch to our docs later this week to update our usage docs to document some of expections we have more explictly
15:02:05 sean-k-mooney jangutter: i useally add [os-vif] and [nova][neutron]
15:02:12 sean-k-mooney but not really
15:14:19 openstackgerrit Balazs Gibizer proposed openstack/nova master: Recalculate request group - RP mapping during re-schedule https://review.openstack.org/619529
15:14:19 openstackgerrit Balazs Gibizer proposed openstack/nova master: Send RP uuid in the port binding https://review.openstack.org/569459
15:14:20 openstackgerrit Balazs Gibizer proposed openstack/nova master: Test boot with more ports with bandwidth request https://review.openstack.org/573317
15:14:20 openstackgerrit Balazs Gibizer proposed openstack/nova master: Reject interface attach with QoS aware port https://review.openstack.org/570078
15:14:21 openstackgerrit Balazs Gibizer proposed openstack/nova master: Reject networks with QoS policy https://review.openstack.org/570079
15:18:12 jangutter sean-k-mooney, jaypipes: if I don't need to jump through hoops, I hope https://review.openstack.org/#/c/572081 is a good representative of 'Option 1'.
15:21:03 mriedem dansmith: online data migration pattern check here https://review.openstack.org/#/c/613499/12/nova/objects/compute_node.py@211 - it works, but not sure if it's the "right way"
15:21:21 openstackgerrit Balazs Gibizer proposed openstack/nova master: Deprecate the unversioned notifications https://review.openstack.org/603079
15:21:29 dansmith mriedem: queued
15:21:45 adrianc Ah i really need to setup an IRC bouncer, sean-k-mooney: understood, ive dealt with the mixed version in the code. we will see how it goes.
15:22:39 sean-k-mooney adrianc: i just set up znc on kubernets over the weekend. k8s is a pain on the ass but i now have a bouncer
15:23:48 sean-k-mooney it was so much more work then i had planned vs plain docker
15:23:50 adrianc i may go straight with docker then :)
15:25:33 sean-k-mooney well i had to set up helm to deploy nginx/ingress contoller + certmanger for lets encyrpet certs to then deploy znc with config via config maps + an init sidecar container
15:25:56 sean-k-mooney or i could have deployed 2 contienrs by hand.
15:26:34 sean-k-mooney my k8s cluster is a single node on an old laptop so its not like ita actully does anyting other then booting a docker container anyway
15:49:34 adrianc thinking about it further maybe ill spin up a VM on my development machine and run it there, quicker ramp-up time as i havent actually used docker but it may be a good experience on the other hand :)
15:50:32 sean-k-mooney you can always just install znc to the host. people are still allowed to do that :)
15:51:10 cdent heresy!
15:54:57 adrianc lol
15:55:31 cdent mriedem: on the nova-status tests: they rely on the placement database. Would you prefer: a) move the test into the functional hierarchy so only functional tests are using the "remote" placement database fixture, b) unwind the removal of a "local" placement database fixture (in nova/test.py, c) mock the shit out of that stuff, or d) not bother and leave the test removed?
16:00:52 cdent efried, jaypipes, dansmith, edleafe ^
16:02:50 cdent e) use the placement database fixture under 'unit' is not an option, because only the functional tests install placement
16:03:45 edleafe You know I despise unit tests that test more than a unit
16:04:20 edleafe If it stays in unit, mock it. But moving to functional sounds like a better option
16:07:12 cdent thanks edleafe, for now I'm going with move to functional as anything else is way more work than it is worth
16:07:59 jaypipes cdent: my vote would be same as Ed's. move it to functional...
16:08:34 cdent thanks jaypipes
16:09:13 gibi +1 for moving it to functional :)
16:15:11 mriedem sorry was in another channel,
16:15:32 mriedem or e) i could rewrite those tests to use the placement rest api fixture (not the db fixture) in a patch below this one
16:17:26 cdent mriedem: the test, as currently written in master, cannot pass with the other changes in the patch, which remove the placement database, locally
16:17:53 cdent so it would definitely need to be a prioer to this one, if you're feeling inclined
16:18:24 mriedem but the placement fixture (rest api) continues to work, right? but it gets the fixture from placement.
16:19:27 cdent yes, but that fixture is only available to functional tests, not unit
16:19:43 cdent (because only the functional tox jobs import placement master)
16:20:01 cdent this seemed a good safeguard to insure that unit tests are unit tests
16:20:11 cdent and is part of why this test ran into issues
16:21:03 mriedem ok in that case i guess (a) is best for now, unless we just drop the check altogether as i think we sort of flirted with last week
16:21:59 cdent roger that, I'll carry on, thanks
16:26:41 mriedem ok, i'm going to dump some words into the ML
16:42:04 edleafe Must have been a big dump
16:43:38 dansmith I call BS on the 5lb number
16:43:38 dansmith https://www.healthline.com/health/do-you-lose-weight-when-you-poop
16:45:21 artom I think BS (from bulls!) is definitely more than 5 lbs.
16:45:48 dansmith probably, but mriedem is decidedly smaller than a bull
16:46:47 bauzas are we discussing how our body filters and weights stuff ?
16:47:30 mriedem i'm going to take the high road here
16:47:35 mriedem just this once
16:48:59 artom My son blocked the toilet with his poop. Twice. He's not even 3 yet.
17:00:43 mriedem i've seen that from a kid once, at my gf's house in high school, her nephew birthed a loaf of bread i swear, it was amazing
17:00:57 cdent so much for the high road
17:01:04 mriedem you can thank artom
17:01:07 mriedem he brings out the worst in me
17:01:20 cdent yeah, he just gently laid that in there, and the house came down
17:04:32 dansmith artom: that's quite an, um, honor
17:08:06 artom I did to this channel what my son did to the toilet
17:08:20 artom I am... this channel's poop? Dammit >_<
17:08:39 dansmith we can't even see the high road from here
17:10:30 artom Yeah, but pot's legal in Canada now, so if I ever need to get to the high road, I'm all set
17:10:53 artom Which is why I can afford these shenanigans
17:22:03 openstackgerrit Jack Ding proposed openstack/nova master: [WIP] Preserve UEFI NVRAM variable store https://review.openstack.org/621646
17:22:16 dansmith artom: when are you thinking you'll have numa-aware-live-migration patches up/refreshed for looking?
17:32:30 openstackgerrit Merged openstack/nova master: Fix sloppy initialization of the new disk ops semaphore. https://review.openstack.org/621346
17:32:47 jaypipes sean-k-mooney: I think we need mdulko's response on your ML thread.
17:34:44 artom dansmith, it's in my tasks for this downstream sprint
17:34:53 artom So I'm not sure how much I'll get done, but at least 1
17:36:06 artom dansmith, I'm not really expecting it to land by our downstream deadlines. But if I can get it in upstream in Stein, and make it tech preview for OSP15, I'd consider that a win
17:36:54 dansmith artom: okay, in the upstream channel I'm really only asking about things that matter to upstream :)
17:37:09 sean-k-mooney jaypipes: is there a responce to the os-vif one or is it another one?
17:37:14 dansmith just wondering if there will be meat to review this year or not, as it doesn't seem like there's much up right now
17:37:29 artom dansmith, I know, but was giving you context
17:37:55 dansmith alright
17:38:03 artom dansmith, maybe not a full 15 oz steak, but like, a chicken cutlet?
17:39:42 jaypipes sean-k-mooney: no, I mean you are asking for kuryr-kubernetes folks to respond to your questions in your ML thread and I'm saying I'll wait to hear from them until I weigh in.
17:39:57 sean-k-mooney jaypipes: ah ok
17:40:54 sean-k-mooney ya ill proably jump into there irc tommorw and flag it to them
17:54:08 mriedem so we're not trying hard to get numa-aware live migration done in stein upstream, got it
17:58:12 efried jaypipes: I redid the three patches starting at https://review.openstack.org/#/c/620711/ per your comments. Gate failures are spurious, rechecking. tssurya confirmed they've been running with this code in their env and it's performing as expected.
17:58:19 openstackgerrit Matt Riedemann proposed openstack/nova master: Add functional regression test for bug 1806064 https://review.openstack.org/621664
17:58:19 openstack bug 1806064 in OpenStack Compute (nova) rocky "Volume remains in attaching/reserved status, if the instance is deleted after TooManyInstances exception in nova-conductor" [Medium,Triaged] https://launchpad.net/bugs/1806064
17:58:19 sean-k-mooney mriedem: it will likely be hard to get it runway ready and reviewd before jan 10, it likely can be dont before upstream feature freeze but it likely will be non trival to review
17:58:20 mriedem s10: melwitt: ^
17:58:45 cfriesen artom: we'll happily try to break any patches you come up with. :)
17:58:54 mriedem sean-k-mooney: feature freeze is not jan 10
17:59:29 mriedem FF is march 7
18:00:29 sean-k-mooney oh i taught it was mid febuary but i guess im mixing that up with non-client libs
18:01:09 tssurya efried: we will try to put up a blog post soon with the plots and perf-improvement
18:01:35 efried tssurya: That would be awesome, my boss was just asking me if such a thing was available.
18:01:41 mriedem tssurya: are there plans to test that out with the ironic computes?
18:01:45 mriedem so those can be upgraded?
18:02:08 mriedem you had to downgrade ironic computes to queens right?
18:02:14 tssurya mriedem: yea we are working on the ironic driver and for sure will have it upgraded asap
18:02:27 tssurya mriedem: yea the prod is still in queens for ironic compute
18:02:59 mriedem efried: do i smell pbc results?
18:03:06 mriedem er,

Earlier   Later