| *** ttsiouts has quit IRC | 00:06 | |
| *** takashin has joined #openstack-placement | 00:07 | |
| openstackgerrit | Chris Dent proposed openstack/placement master: WIP: Remove null root provider protections https://review.opendev.org/657716 | 01:39 |
|---|---|---|
| *** cdent has quit IRC | 01:40 | |
| *** tssurya has quit IRC | 02:16 | |
| *** ttsiouts has joined #openstack-placement | 02:39 | |
| *** ttsiouts has quit IRC | 02:44 | |
| *** cdent has joined #openstack-placement | 04:34 | |
| *** cdent has quit IRC | 05:09 | |
| *** dims has quit IRC | 05:32 | |
| *** dims has joined #openstack-placement | 05:33 | |
| *** dims has quit IRC | 06:16 | |
| *** ttsiouts has joined #openstack-placement | 06:34 | |
| *** dims has joined #openstack-placement | 06:42 | |
| *** dims has quit IRC | 06:49 | |
| *** dims has joined #openstack-placement | 06:54 | |
| *** ttsiouts has quit IRC | 07:33 | |
| *** takashin has left #openstack-placement | 08:00 | |
| *** helenafm has joined #openstack-placement | 09:06 | |
| *** ttsiouts has joined #openstack-placement | 09:06 | |
| *** ttsiouts has quit IRC | 09:11 | |
| *** ttsiouts has joined #openstack-placement | 10:46 | |
| *** ttsiouts has quit IRC | 10:50 | |
| *** belmoreira has joined #openstack-placement | 11:02 | |
| *** tssurya has joined #openstack-placement | 11:02 | |
| *** sean-k-mooney has joined #openstack-placement | 11:18 | |
| *** belmoreira has quit IRC | 11:36 | |
| *** belmoreira has joined #openstack-placement | 11:39 | |
| *** helenafm has quit IRC | 12:05 | |
| *** ttsiouts has joined #openstack-placement | 12:32 | |
| *** alex_xu has quit IRC | 12:43 | |
| *** alex_xu has joined #openstack-placement | 12:48 | |
| *** ttsiouts has quit IRC | 12:53 | |
| *** alex_xu has quit IRC | 13:01 | |
| efried_pto | stephenfin: Can you suss why https://review.opendev.org/#/c/655665/1 is failing the docs job? | 13:12 |
| *** efried_pto is now known as efried | 13:12 | |
| stephenfin | efried: looks like os-resource-classes haven't specifically stated that python3 be used to build docs http://logs.openstack.org/65/655665/1/check/openstack-tox-docs/9ed3570/job-output.txt.gz#_2019-05-08_12_57_27_563476 | 13:13 |
| stephenfin | i.e. 'basepython = python3' | 13:13 |
| stephenfin | in tox.ini | 13:13 |
| efried | stephenfin: Okay, so does it depend which CI node it lands on for the job or did something change since Monday? | 13:14 |
| efried | Because https://review.opendev.org/#/c/657464/ passed | 13:14 |
| stephenfin | efried: I'm guessing something changed under the hood, though what I don't know | 13:15 |
| efried | I can propose a patch on tox.ini I guess... | 13:16 |
| stephenfin | because Sphinx 2.0 is Python 3 only so that shouldn't ever work with Python 2.7 | 13:16 |
| efried | yah | 13:16 |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Use py3 for docs build https://review.opendev.org/657800 | 13:17 |
| efried | stephenfin: ^ | 13:17 |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Add MEM_ENCRYPTION_CONTEXT resource class https://review.opendev.org/655665 | 13:17 |
| efried | thanks stephenfin, will see how this shakes out. | 13:18 |
| efried | stephenfin: Yeah, https://review.opendev.org/#/c/657464/ failed today, so something must have changed somewhere since Monday night. | 13:30 |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Propose standard ACCELERATOR_FPGA resource class https://review.opendev.org/657464 | 13:30 |
| *** ttsiouts has joined #openstack-placement | 13:30 | |
| efried | stephenfin, gibi: Would y'all mind approving orc gate fix https://review.opendev.org/#/c/657800/ please | 13:32 |
| stephenfin | efried: Before we do, do you want to fix all the other targets? Everything should be using python3 now | 13:35 |
| stephenfin | efried: i.e. put it in '[testenv]' and set the ignore_basepython_conflict option in '[tox]' | 13:35 |
| efried | stephenfin: Except for the py27 env. That one will ignore the testenv setting, yah? | 13:37 |
| stephenfin | efried: If you set ignore_basepython_conflict, yeah | 13:37 |
| stephenfin | Otherwise tox will complain | 13:37 |
| *** helenafm has joined #openstack-placement | 13:38 | |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Use py3 for all* tox targets https://review.opendev.org/657800 | 13:39 |
| efried | stephenfin: Like that ^ | 13:39 |
| stephenfin | yup | 13:41 |
| *** jaypipes has joined #openstack-placement | 13:43 | |
| *** aspiers has joined #openstack-placement | 13:46 | |
| aspiers | o/ | 13:46 |
| aspiers | I still don't get that sphinx failure | 13:51 |
| aspiers | When I run tox -e docs on os-resource-classes locally, it installs Sphinx 1.8.5 | 13:51 |
| *** mriedem has joined #openstack-placement | 13:52 | |
| aspiers | which works fine with Python 2.7 | 13:52 |
| aspiers | Whereas http://logs.openstack.org/65/655665/1/gate/openstack-tox-docs/6da0012/job-output.txt.gz installed 2.0.1 which failed with 2.7.15 | 13:52 |
| *** artom has joined #openstack-placement | 13:52 | |
| aspiers | doc/requirements.txt has sphinx!=1.6.6,!=1.6.7,>=1.6.2 # BSD | 13:52 |
| aspiers | so is this just pip being non-deterministic? | 13:53 |
| aspiers | stephenfin, efried: ^^^ | 13:53 |
| stephenfin | aspiers: You tried 'tox -e docs --recreate' ? | 13:54 |
| aspiers | I nuked .tox/docs first | 13:54 |
| aspiers | which is basically the same thing IIUC | 13:54 |
| aspiers | I guess pip is too stupid to pick the right Sphinx version based on the Python version, but that still doesn't explain why it works for me locally | 13:55 |
| * aspiers tries with --recreate just to be sure | 13:56 | |
| aspiers | yup, still works | 13:56 |
| aspiers | Installs Sphinx 1.8.5 | 13:57 |
| stephenfin | aspiers: Yeah, looking at 'openstack/requirements', I see the following: | 13:57 |
| stephenfin | 413 | Sphinx===1.8.5;python_version=='2.7' | 13:57 |
| stephenfin | 413 | Sphinx===1.8.5;python_version=='2.7' | 13:57 |
| stephenfin | 414 | Sphinx===2.0.1;python_version=='3.4' | 13:58 |
| aspiers | Ah, so pip *isn't* that stupid | 13:58 |
| stephenfin | last two for python_version==('3.5'|'3.6') too | 13:58 |
| aspiers | in which case, why the hell is the CI job installing Sphinx 2.x? | 13:58 |
| stephenfin | not a clue. That's a weird one | 13:58 |
| aspiers | Is the CI job actually using the requirements repo? I see it being cloned, but not used | 13:59 |
| stephenfin | aspiers: Ah, indeed | 13:59 |
| stephenfin | We're not using upper-constraints | 13:59 |
| aspiers | it's in tox.ini | 13:59 |
| aspiers | but overridden in [testenv:docs] | 14:00 |
| stephenfin | for '[testenv] deps', yes | 14:00 |
| stephenfin | yeah, not for the docs version | 14:00 |
| aspiers | that's the bug | 14:00 |
| stephenfin | yuuup | 14:00 |
| aspiers | so we shouldn't need to require Python 2.7 | 14:00 |
| aspiers | but fix this instead | 14:00 |
| aspiers | sorry | 14:00 |
| aspiers | s/2.7/3/ | 14:00 |
| stephenfin | Indeed. Nice spot | 14:00 |
| aspiers | Well, you gave me the right hint by mentioning openstack/requirements ;-) | 14:01 |
| stephenfin | We still should switch to Python3 but this is the main issue | 14:01 |
| aspiers | Right | 14:01 |
| aspiers | I can submit the fix | 14:01 |
| stephenfin | wanna propose the fix? | 14:01 |
| aspiers | :) | 14:01 |
| stephenfin | jinx | 14:01 |
| stephenfin | got for it | 14:01 |
| aspiers | OK, I guess just by duplicating the upper-constraints line into testenv:docs? | 14:02 |
| stephenfin | yup | 14:02 |
| stephenfin | interestingly, nova is also broken in this way | 14:02 |
| aspiers | ooohh | 14:02 |
| stephenfin | actually, no, it's not | 14:02 |
| aspiers | :) | 14:02 |
| stephenfin | because we override install_command there instead | 14:02 |
| aspiers | os-traits is correct | 14:03 |
| stephenfin | though mordred said there's some reason not to do that. idk why | 14:03 |
| aspiers | stephenfin: It still doesn't explain why the CI job chose Sphinx 2.x but my local tox/pip didn't | 14:05 |
| aspiers | so I can't reproduce the issue locally :-/ | 14:06 |
| aspiers | ah well, I'll submit anyway | 14:06 |
| stephenfin | I wonder if they're using the a different Python version of the tox package itself | 14:06 |
| aspiers | Yeah, or pip | 14:06 |
| stephenfin | s/the // | 14:06 |
| stephenfin | yup | 14:06 |
| stephenfin | I'd rather it did fail everywhere, tbh. At least that way we'd spot the breakages | 14:07 |
| aspiers | Right :) | 14:07 |
| *** belmoreira has quit IRC | 14:08 | |
| openstackgerrit | Adam Spiers proposed openstack/os-resource-classes master: Add missing upper-constraints.txt to deps in docs testenv https://review.opendev.org/657815 | 14:13 |
| aspiers | stephenfin: ^^^ | 14:13 |
| aspiers | OK, back in ~15 | 14:13 |
| aspiers | efried: ^^^ | 14:13 |
| efried | aspiers: +A | 14:25 |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Add MEM_ENCRYPTION_CONTEXT resource class https://review.opendev.org/655665 | 14:26 |
| efried | mriedem: Mind pushing https://review.opendev.org/#/c/657800/ please? | 14:27 |
| mriedem | looking | 14:29 |
| mriedem | done | 14:31 |
| openstackgerrit | Merged openstack/os-resource-classes master: Add missing upper-constraints.txt to deps in docs testenv https://review.opendev.org/657815 | 14:37 |
| aspiers | \o/ | 14:37 |
| openstackgerrit | Merged openstack/os-resource-classes master: Use py3 for all* tox targets https://review.opendev.org/657800 | 14:41 |
| openstackgerrit | Merged openstack/os-resource-classes master: Add MEM_ENCRYPTION_CONTEXT resource class https://review.opendev.org/655665 | 14:47 |
| *** belmoreira has joined #openstack-placement | 14:49 | |
| *** artom has quit IRC | 14:50 | |
| *** artom has joined #openstack-placement | 14:51 | |
| openstackgerrit | Eric Fried proposed openstack/os-resource-classes master: Propose standard ACCELERATOR_FPGA resource class https://review.opendev.org/657464 | 14:52 |
| *** belmoreira has quit IRC | 15:07 | |
| *** ttsiouts has quit IRC | 15:07 | |
| *** ttsiouts has joined #openstack-placement | 15:12 | |
| *** Sundar has joined #openstack-placement | 15:19 | |
| *** altlogbot_0 has quit IRC | 15:21 | |
| *** irclogbot_2 has quit IRC | 15:22 | |
| *** altlogbot_0 has joined #openstack-placement | 15:22 | |
| *** irclogbot_0 has joined #openstack-placement | 15:24 | |
| *** altlogbot_0 has quit IRC | 15:27 | |
| *** altlogbot_0 has joined #openstack-placement | 15:28 | |
| *** irclogbot_0 has quit IRC | 15:28 | |
| *** irclogbot_2 has joined #openstack-placement | 15:30 | |
| *** cdent has joined #openstack-placement | 15:36 | |
| *** e0ne has joined #openstack-placement | 15:50 | |
| *** e0ne has quit IRC | 15:53 | |
| cdent | mriedem: do you have any more info than what's currently on https://review.opendev.org/#/c/656448/ about what's up with watcher? | 16:06 |
| mriedem | not yet no | 16:08 |
| mriedem | and i don't know enough about the watcher internals yet to be able to identify things they are doing with nova that could be replaced with placement | 16:08 |
| mriedem | my guess is they could use GET /a_c as a pre-check to validate a set of nodes before asking nova to migrate somewhere | 16:09 |
| mriedem | it's also not clear to me how watcher gathers metrics for inventory and allocation data it uses to make decisions | 16:10 |
| mriedem | i mean, i know they have gnocchi/ceilometer/monasca data sources so i'm guessing they rely on those things listening for that info from nova, and then based on that make their decisions | 16:10 |
| *** e0ne has joined #openstack-placement | 16:11 | |
| *** artom has quit IRC | 16:18 | |
| mriedem | i've just been told that we need to use watcher (or write our own - which i'm trying to really avoid) but we need to make sure it scales for a public cloud | 16:19 |
| mriedem | so i'm kind of fumbling in the dark, like a 6th grader at a lock in | 16:19 |
| *** david-lyle has quit IRC | 16:19 | |
| *** dklyle has joined #openstack-placement | 16:20 | |
| cdent | mriedem: that sounds like great type ii fun | 16:24 |
| *** helenafm has quit IRC | 16:24 | |
| *** e0ne has quit IRC | 16:26 | |
| mriedem | i have a decent story about a 6th grade lock in where i was trying to woo a girl and ended the night hunched over a toilet in the gym locker room | 16:26 |
| mriedem | ahhh youth | 16:26 |
| aspiers | XD | 16:27 |
| aspiers | Sounds about as successful as some of my early attempts | 16:27 |
| mriedem | oh i still "dated" angie for awhile after that night, my powers of charisma could not be stopped - until some time later when i'm sure i was broken up with via one of her friends telling me, or a note or something | 16:29 |
| aspiers | life is harsh | 16:31 |
| edleafe | mriedem: The problem Watcher has (or at least *had* when I was working with them) was that even if they knew that a target host had sufficient resources, they had no visibility into the restrictions that were in scheduler filters | 17:00 |
| edleafe | E.g., affinity/anti | 17:00 |
| edleafe | That was why their ask has been for a way to run potential candidates through the scheduler to get a filtered set that passes whatever restrictions are in place | 17:01 |
| *** ttsiouts has quit IRC | 17:10 | |
| *** e0ne has joined #openstack-placement | 17:10 | |
| *** e0ne has quit IRC | 17:12 | |
| *** e0ne has joined #openstack-placement | 17:18 | |
| mriedem | yeah i thought about that, plus any qualitative things now like traits, and we have these pre-filters for aggregates and such | 17:18 |
| *** ttsiouts has joined #openstack-placement | 17:19 | |
| *** e0ne has quit IRC | 17:21 | |
| edleafe | mriedem: Since the placement API is accessible to them, they could handle those queries. The Nova scheduler, though, has no external access | 17:21 |
| *** ttsiouts has quit IRC | 17:54 | |
| *** e0ne has joined #openstack-placement | 17:59 | |
| *** e0ne has quit IRC | 18:07 | |
| *** ttsiouts has joined #openstack-placement | 18:09 | |
| *** artom has joined #openstack-placement | 18:10 | |
| *** ttsiouts has quit IRC | 18:14 | |
| *** ttsiouts has joined #openstack-placement | 18:42 | |
| *** cdent has quit IRC | 19:20 | |
| *** ttsiouts has quit IRC | 19:44 | |
| *** ttsiouts has joined #openstack-placement | 19:59 | |
| *** cdent has joined #openstack-placement | 20:03 | |
| *** ttsiouts has quit IRC | 20:04 | |
| *** david-lyle has joined #openstack-placement | 20:12 | |
| *** dklyle has quit IRC | 20:15 | |
| *** edleafe has quit IRC | 20:15 | |
| *** evrardjp has quit IRC | 20:15 | |
| *** dansmith has quit IRC | 20:15 | |
| *** edleafe has joined #openstack-placement | 20:16 | |
| *** evrardjp has joined #openstack-placement | 20:16 | |
| *** purplerbot has quit IRC | 20:17 | |
| *** dansmith has joined #openstack-placement | 20:20 | |
| *** ttsiouts has joined #openstack-placement | 20:37 | |
| *** purplerbot has joined #openstack-placement | 20:40 | |
| *** tssurya has quit IRC | 20:41 | |
| *** ttsiouts has quit IRC | 20:42 | |
| *** artom has quit IRC | 20:44 | |
| *** artom has joined #openstack-placement | 20:45 | |
| *** artom has quit IRC | 20:45 | |
| *** cdent has quit IRC | 20:55 | |
| *** cdent has joined #openstack-placement | 21:00 | |
| *** cdent has quit IRC | 21:08 | |
| *** cdent has joined #openstack-placement | 21:33 | |
| openstackgerrit | Eric Fried proposed openstack/os-traits master: Introduce traits for NUMA and NIC subtree affinity https://review.opendev.org/657898 | 22:09 |
| efried | cdent: Created a Task for ^, not sure if we care to make tasks that granular or not... | 22:10 |
| efried | uhh, I guess I should have created the Task under os-traits? | 22:10 |
| efried | Not sure how that works. | 22:10 |
| cdent | if it helps get things done, that's great | 22:10 |
| cdent | when you can have tasks in other projects under a story that is in project A | 22:10 |
| efried | okay, I was able to change the project for that task. | 22:11 |
| cdent | btw, I create a worklist with nested magic tasks on it: https://storyboard.openstack.org/#!/worklist/627 | 22:11 |
| * efried looks to see if new task magically ended up in same... | 22:11 | |
| efried | yup, did it end up there because you put it there or did it happen automatically cdent? | 22:11 |
| cdent | automatically | 22:11 |
| efried | neat | 22:12 |
| efried | so, let the bikeshedding begin. | 22:12 |
| cdent | those are "active tasks on story ####" | 22:12 |
| cdent | which was far harder to suss out than it should have been | 22:12 |
| cdent | the deal on those traits is that every numa node and every nic would get them? | 22:14 |
| cdent | i guess automating that is why we have computesr | 22:14 |
| * cdent gack. i am so tired. | 22:14 | |
| efried | yeah, that's what I'm thinking | 22:20 |
| efried | it seems like a reasonable use of a trait, to me. | 22:20 |
| efried | again, where "trait" means "trait", in the sense of "characteristic", not necessarily "capability". | 22:21 |
| * cdent nods | 22:22 | |
| *** cdent has quit IRC | 23:15 | |
| openstackgerrit | Eric Fried proposed openstack/placement master: Add NUMANetworkFixture for gabbits https://review.opendev.org/657463 | 23:19 |
| *** e0ne has joined #openstack-placement | 23:35 | |
| *** ttsiouts has joined #openstack-placement | 23:41 | |
| *** cdent has joined #openstack-placement | 23:44 | |
| *** e0ne has quit IRC | 23:52 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!