| openstackgerrit | Sundar Nadathur proposed openstack/nova-specs master: Nova Cyborg interaction specification. https://review.openstack.org/603955 | 00:46 |
|---|---|---|
| *** tetsuro has joined #openstack-placement | 01:05 | |
| *** tetsuro has quit IRC | 01:41 | |
| *** tetsuro has joined #openstack-placement | 01:41 | |
| *** tetsuro has quit IRC | 01:51 | |
| *** takashin has quit IRC | 01:52 | |
| *** tetsuro has joined #openstack-placement | 01:55 | |
| *** tetsuro has quit IRC | 02:01 | |
| *** tetsuro has joined #openstack-placement | 02:03 | |
| *** takashin has joined #openstack-placement | 02:36 | |
| openstackgerrit | Sundar Nadathur proposed openstack/nova-specs master: Nova Cyborg interaction specification. https://review.openstack.org/603955 | 03:09 |
| *** takashin has quit IRC | 03:30 | |
| *** takashin has joined #openstack-placement | 03:55 | |
| *** tetsuro has quit IRC | 04:06 | |
| *** tetsuro has joined #openstack-placement | 05:14 | |
| *** ttsiouts has joined #openstack-placement | 06:59 | |
| *** ttsiouts has quit IRC | 07:16 | |
| *** helenafm has joined #openstack-placement | 07:26 | |
| *** ttsiouts has joined #openstack-placement | 08:00 | |
| *** tssurya has joined #openstack-placement | 08:03 | |
| *** e0ne has joined #openstack-placement | 08:42 | |
| *** ttsiouts has quit IRC | 08:49 | |
| *** ttsiouts has joined #openstack-placement | 08:59 | |
| *** ttsiouts has quit IRC | 09:04 | |
| *** ttsiouts has joined #openstack-placement | 09:08 | |
| *** s10 has joined #openstack-placement | 09:26 | |
| *** ttsiouts has quit IRC | 10:21 | |
| *** ttsiouts has joined #openstack-placement | 10:31 | |
| *** cdent has joined #openstack-placement | 11:14 | |
| *** ttsiouts has quit IRC | 11:32 | |
| *** tetsuro has quit IRC | 12:04 | |
| *** ttsiouts has joined #openstack-placement | 12:07 | |
| jaypipes | cdent: quick question on this: <<: *cn1_zero | 12:14 |
| jaypipes | cdent: can that go in the first line of a YAML node? | 12:14 |
| cdent | jaypipes: aye aye? | 12:14 |
| jaypipes | cdent: that makes the node inherit from an anchor, right? | 12:14 |
| cdent | yeah, i've changed it in the latest version | 12:15 |
| jaypipes | cdent: ah, yes, I see you have. | 12:15 |
| jaypipes | cdent: question answered then :) | 12:15 |
| jaypipes | cdent: back to your regularly scheduled programming :) | 12:16 |
| cdent | high latency grenade development | 12:16 |
| *** tetsuro has joined #openstack-placement | 13:25 | |
| cdent | one more bug, one more tweak, one more wait | 13:44 |
| efried | gibi (and jaypipes): Consumer generations aren't returned in GET /a_c, right? Because how could they be? | 13:53 |
| gibi | efried: right | 13:53 |
| efried | Which means that in the scheduler flow, we actually have to *add* a consumer generation field to the allocation request we pull out of the GET /a_c return | 13:54 |
| cdent | yes | 13:54 |
| efried | Which means we're no longer in the happy place where that payload is opaque | 13:54 |
| cdent | we were already there | 13:54 |
| cdent | because we also add a project id and user id | 13:54 |
| efried | Like, we knew we had to introspect for weighers and late filters | 13:54 |
| efried | oh, that wasn't there before either, yeah that makes sense. | 13:55 |
| efried | Though I guess if we ever implement quotas we'll probably send proj/user in the GET /a_c request and it could be returned... | 13:55 |
| gibi | for the scheduler perspective every claim is a new consumer (except non-forced evacuate :/) | 13:55 |
| efried | Okay, so I guess the line has moved, but we're still able to say that the "allocations" member of the allocation request is opaque | 13:56 |
| gibi | efried: yes, that is not touched (except evacuate) | 13:56 |
| efried | Yeah, gibi, so to that point, I assume we're eventually going to move toward evacuate also using migration UUID? | 13:57 |
| gibi | efried: I have a todo from the PTG to do that move | 13:57 |
| efried | And we haven't yet because, what, it was too hairy of a problem to deal with when we did live migrations? | 13:57 |
| efried | So at that point, *every* PUT /allocations/{c} coming from a GET /a_c response ought to have consumer_generation=None? | 13:57 |
| gibi | efried: let me double-check | 13:58 |
| gibi | efried: if you mean claim_resources instead of PUT allocations/{c} then I think it is true | 13:58 |
| efried | so like, I'm wondering if it makes sense for GET /a_c to include "consumer_generation": null in the response. | 13:59 |
| efried | And as soon as I say it, I realize the answer is no | 13:59 |
| efried | but | 13:59 |
| efried | if we had an actual consumer endpoint | 13:59 |
| efried | then we would have an operation PUT /allocations where the consumer (with proj/user IDs and generation, I think) was part of the payload, not a UUID in the URI. | 14:00 |
| efried | This may be what jaypipes has been getting at. | 14:00 |
| efried | and then the GET /a_c response unit *could* be truly opaque. | 14:01 |
| efried | (except for some cases of weighers and late filters, maybe, not sure) | 14:01 |
| gibi | efried: but we have https://developer.openstack.org/api-ref/placement/#manage-allocations | 14:01 |
| gibi | efried: there the consumer is in the payload | 14:01 |
| efried | gibi: Yeah, I think I'm saying that would be kind of restructured | 14:02 |
| cdent | I kind of prefer it that a GET /a_c is not "user" aware | 14:02 |
| cdent | at least until it is absolutely necessary for it to be so | 14:02 |
| efried | Yes. But a PUT is | 14:02 |
| efried | So you GET /a_c and it returns allocations only | 14:02 |
| efried | then you take that dict, blindly, and stuff it *next to* a dict describing the consumer to do your PUT /allocations (or a new version of POST as gibi says) | 14:03 |
| jaypipes | efried, cdent: it was always about the structure/format of the allocation_request part of the GET /a_c response. That is what we care about versioning which is why the version supplied to claim_resources() is called allocation_request_version. | 14:06 |
| *** s10_ has joined #openstack-placement | 14:28 | |
| *** s10 has quit IRC | 14:30 | |
| *** cdent has quit IRC | 14:53 | |
| *** e0ne has quit IRC | 14:58 | |
| *** takashin has left #openstack-placement | 15:01 | |
| *** cdent has joined #openstack-placement | 15:03 | |
| *** s10_ has quit IRC | 15:09 | |
| *** ttsiouts has quit IRC | 15:11 | |
| *** e0ne has joined #openstack-placement | 15:13 | |
| openstackgerrit | Eric Fried proposed openstack/nova-specs master: WIP: High Precision Event Timer (HPET) on x86 guests https://review.openstack.org/607989 | 15:24 |
| *** e0ne has quit IRC | 15:47 | |
| *** helenafm has quit IRC | 16:03 | |
| edleafe | cdent: ok, I blew away my failing devstack VM and created a fresh one. It's working, except that it didn't install placement. Is https://review.openstack.org/#/c/600162 the only patch I need? | 16:20 |
| cdent | edleafe: yes | 16:20 |
| edleafe | cdent: ty | 16:21 |
| *** tssurya has quit IRC | 17:48 | |
| jaypipes | gibi: rock on. consumer gen merged. | 18:29 |
| jaypipes | gibi: so what was your expectation with regards to getting through the remaining patches in this series? today? | 18:30 |
| cdent | praise be | 18:30 |
| jaypipes | lord have mercy | 18:30 |
| *** e0ne has joined #openstack-placement | 18:31 | |
| cdent | I _think_ i'm 1, maybe 2, iterations from having the grenade stuff working | 18:31 |
| cdent | at which point I'll feel happy about taking next week off | 18:32 |
| *** e0ne has quit IRC | 18:33 | |
| jaypipes | cdent: nice work on that. | 18:33 |
| cdent | mriedem laid in most of the ground work, and then I found all little niggly bits | 18:34 |
| *** s10 has joined #openstack-placement | 18:34 | |
| *** mriedem has joined #openstack-placement | 18:34 | |
| cdent | i'll try to write a reasonable summary in tomorrow pupdate so there's no state up in the air | 18:34 |
| cdent | speak of the devil | 18:34 |
| *** cdent has quit IRC | 18:39 | |
| efried | jaypipes: FYI I'm working my way up the stack, working through https://review.openstack.org/#/c/605785 atm | 19:08 |
| efried | (regarding gibi's series) | 19:08 |
| *** mriedem has quit IRC | 19:15 | |
| *** s10 has quit IRC | 19:17 | |
| jaypipes | efried: k, I'm still chugging through em as well. | 19:19 |
| *** cdent has joined #openstack-placement | 19:23 | |
| *** mriedem has joined #openstack-placement | 19:24 | |
| *** s10 has joined #openstack-placement | 19:24 | |
| *** cdent has quit IRC | 19:54 | |
| *** e0ne has joined #openstack-placement | 20:08 | |
| *** e0ne has quit IRC | 20:14 | |
| *** s10 has quit IRC | 20:16 | |
| *** s10 has joined #openstack-placement | 20:16 | |
| *** s10 has quit IRC | 20:16 | |
| *** s10 has joined #openstack-placement | 20:17 | |
| *** s10 has quit IRC | 20:17 | |
| *** s10 has joined #openstack-placement | 20:18 | |
| *** s10 has quit IRC | 20:18 | |
| *** s10 has joined #openstack-placement | 20:18 | |
| *** s10 has quit IRC | 20:19 | |
| *** s10 has joined #openstack-placement | 20:19 | |
| *** s10 has quit IRC | 20:20 | |
| *** e0ne has joined #openstack-placement | 20:30 | |
| *** e0ne has quit IRC | 21:00 | |
| *** mriedem has quit IRC | 21:05 | |
| *** edleafe has quit IRC | 23:39 | |
| *** takashin has joined #openstack-placement | 23:48 | |
| *** tetsuro_ has joined #openstack-placement | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!