| *** artom has quit IRC | 03:27 | |
| openstackgerrit | Huachang Wang proposed openstack/os-traits master: CPU: add a trait for AVX512-VNNI support https://review.opendev.org/672888 | 07:02 |
|---|---|---|
| *** ttsiouts has joined #openstack-placement | 08:03 | |
| *** tssurya has joined #openstack-placement | 08:08 | |
| *** tssurya has quit IRC | 08:23 | |
| *** cdent has joined #openstack-placement | 09:06 | |
| *** e0ne has joined #openstack-placement | 09:16 | |
| *** ttsiouts has quit IRC | 09:46 | |
| *** ttsiouts has joined #openstack-placement | 09:47 | |
| *** ttsiouts has quit IRC | 09:51 | |
| *** ttsiouts has joined #openstack-placement | 10:01 | |
| *** ttsiouts has quit IRC | 10:25 | |
| *** ttsiouts has joined #openstack-placement | 10:25 | |
| *** ttsiouts has quit IRC | 10:30 | |
| *** ttsiouts has joined #openstack-placement | 11:06 | |
| *** tssurya has joined #openstack-placement | 11:08 | |
| *** irclogbot_3 has quit IRC | 11:53 | |
| *** irclogbot_0 has joined #openstack-placement | 11:55 | |
| *** artom has joined #openstack-placement | 12:04 | |
| efried | cdent: heads up, once this is ready it'll need a non-Intel core to send it https://review.opendev.org/#/c/672888/ | 12:13 |
| cdent | aye aye | 12:13 |
| *** mriedem has joined #openstack-placement | 12:53 | |
| *** ttsiouts has quit IRC | 12:56 | |
| efried | cdent: I removed my -1 from --^ | 13:05 |
| cdent | aye aye | 13:05 |
| efried | If you would care to render an opinion one way or the other | 13:05 |
| cdent | efried: will get to relative soon, in the midst of a thing | 13:05 |
| efried | I would still prefer the underscore, but the evidence provided makes it less "mandatory" | 13:05 |
| *** artom has quit IRC | 13:09 | |
| *** artom has joined #openstack-placement | 13:09 | |
| *** ttsiouts has joined #openstack-placement | 13:14 | |
| *** artom has quit IRC | 13:26 | |
| efried | cdent: off the top of your head, can request headers be got via environ['HEADERS']? | 13:54 |
| cdent | efried: it's not quite like that | 13:55 |
| cdent | you've got two options | 13:55 |
| efried | glad I asked | 13:55 |
| cdent | if all you have is an environ its environ['HTTP_UPPER_CASE_HEADER_NAME'] | 13:55 |
| cdent | but if you have something like a WebOb Request (req in placement) there 's req.headers which acts like a dict | 13:56 |
| efried | actually, it looks like I have the headers where I need 'em. Thanks. | 13:56 |
| cdent | efried: what you are trying to do? | 13:56 |
| efried | cdent: I noticed this: | 13:57 |
| efried | https://opendev.org/openstack/nova/src/branch/master/nova/scheduler/client/report.py#L547-L556 | 13:57 |
| efried | and wondered why we would need to retrieve the x-openstack-request-id from the response, given that it should just be the same x-openstack-request-id (the global_request_id) we sent in in the first place. | 13:57 |
| efried | But I checked the logs and it ain't | 13:57 |
| efried | http://logs.openstack.org/89/666889/11/check/nova-live-migration/917d052/logs/screen-n-cpu.txt.gz#_Jul_13_12_14_11_461966 | 13:57 |
| efried | That req ID only shows up in that one line on the nova side, and for the one request on the placement side. | 13:57 |
| efried | so I want to throw out some debug patches that dump the headers | 13:58 |
| efried | (btw, same behavior before and after cutting over to the global_request_id kwarg, in case you were wondering) | 13:58 |
| cdent | request id stuff has never seemed to work as advertised | 13:59 |
| cdent | which makes me think there's a concept gap somewhere | 13:59 |
| efried | oy | 14:00 |
| cdent | which is one way of saying: i'm glad you are looking | 14:01 |
| openstackgerrit | Eric Fried proposed openstack/placement master: DNM: Where have all the global_request_id gone? https://review.opendev.org/672984 | 14:05 |
| efried | I have a sneaking suspicion that "%s" % request.headers is going to give me something annoying like "<MultiValuedDict object at 0xaa0789a9b1c234c>" | 14:09 |
| cdent | given how easy it is to run the tests on the placement side, I would have thought you tested that locally first? | 14:11 |
| efried | um | 14:13 |
| efried | why you gotta be all reasonable and stuff? | 14:13 |
| cdent | it's a flaw, I've been working on it | 14:14 |
| efried | hm | 14:15 |
| efried | headers: [('Content-Length', '162'), ('Content-Type', 'application/json'), ('openstack-api-version', 'placement 1.8'), ('vary', 'openstack-api-version')] | 14:17 |
| efried | but when I run with verbose=true I see this coming from... somewhere: | 14:17 |
| efried | < 404 Not Found | 14:17 |
| efried | < Content-Length: 162 | 14:17 |
| efried | < Content-Type: application/json | 14:17 |
| efried | < openstack-api-version: placement 1.8 | 14:17 |
| efried | < vary: openstack-api-version | 14:17 |
| efried | < x-openstack-request-id: req-7f462322-0412-4c63-81aa-e55e8c7180bc | 14:17 |
| cdent | it's probably the order of the stack, lemme check something | 14:18 |
| efried | when I set it up to fail the test, that same ID shows up in the errors list: | 14:18 |
| efried | "errors": [ | 14:18 |
| efried | { | 14:18 |
| efried | "status": 404, | 14:18 |
| efried | "title": "Not Found", | 14:18 |
| efried | "detail": "The resource could not be found.\n\n ", | 14:18 |
| efried | "request_id": "req-7f462322-0412-4c63-81aa-e55e8c7180bc" | 14:18 |
| efried | } | 14:18 |
| efried | ] | 14:18 |
| efried | (404 is the expected behavior btw) | 14:19 |
| efried | back in 5 | 14:21 |
| cdent | yeah, I gotta change locations, will rejoin this in progress when I get to the new place | 14:21 |
| *** cdent has quit IRC | 14:21 | |
| openstackgerrit | Merged openstack/os-traits master: CPU: add a trait for AVX512-VNNI support https://review.opendev.org/672888 | 14:22 |
| *** artom has joined #openstack-placement | 14:26 | |
| *** cdent has joined #openstack-placement | 14:30 | |
| efried | The more os-traits patches I see, the more I dislike the overengineered complexity of it all. We should have just had a straight up list of strings like we do for orc. | 14:42 |
| efried | Do you happen to remember why it was done that way? | 14:42 |
| efried | (I may have asked this before) | 14:42 |
| cdent | efried: it's such a long time ago, I can't remember | 14:48 |
| efried | do you see any benefit today? | 14:48 |
| cdent | if I had to speculate it was something to do with being able to be in particular contexts and a generic love of namespaces | 14:49 |
| cdent | as far as benefit: not really sure, but we're kind of stuck with it because of some guarantees we've made about other code using the library | 14:49 |
| cdent | I was always very conflict about either of the ibraries | 14:50 |
| *** openstackgerrit has quit IRC | 14:51 | |
| cdent | efried: did you learn anything about request ids? I've realized I can't remember what handles the request id injection. i'm guessing the Context | 15:08 |
| efried | cdent: I put those patches up and will get back to their results in a bit. | 15:08 |
| efried | Theoretically: | 15:08 |
| efried | - The global request ID is stored in context.global_id on the nova side | 15:10 |
| efried | - We shove it into the request via a header X-Openstack-Request-Id | 15:10 |
| efried | - It's supposed to be included in the logs by any openstack service on the other end of that request, so you can trace a single "request" (like "create a VM") through all its machinations across various services, like by grepping a unified log file. | 15:10 |
| * cdent nods | 15:13 | |
| cdent | supposed to be | 15:14 |
| *** ttsiouts has quit IRC | 15:27 | |
| *** ttsiouts has joined #openstack-placement | 15:28 | |
| *** ttsiouts has quit IRC | 15:32 | |
| *** ttsiouts has joined #openstack-placement | 15:33 | |
| cdent | latest pupdate: https://anticdent.org/placement-update-19-29.html | 15:41 |
| *** cdent has quit IRC | 16:01 | |
| *** cdent has joined #openstack-placement | 16:06 | |
| cdent | I think I'm done. Any requests before I go? | 16:07 |
| efried | "The two main things on the Placement radar are implementing Consumer | 16:09 |
| efried | Types and cleanups, performance analysis, and documentation related | 16:09 |
| efried | to nested resource providers." <== reverse Rick Perry | 16:09 |
| efried | Have a good evening. | 16:09 |
| * cdent waves | 16:12 | |
| *** e0ne has quit IRC | 16:12 | |
| *** cdent has quit IRC | 16:12 | |
| *** ttsiouts has quit IRC | 16:17 | |
| *** ttsiouts has joined #openstack-placement | 16:17 | |
| *** ttsiouts has quit IRC | 16:22 | |
| *** tssurya has quit IRC | 16:23 | |
| *** mriedem is now known as mriedem_lunch | 16:35 | |
| *** mriedem_lunch is now known as mriedem | 17:50 | |
| *** openstackgerrit has joined #openstack-placement | 17:53 | |
| openstackgerrit | Eric Fried proposed openstack/placement master: Do request_log middleware last https://review.opendev.org/673027 | 17:53 |
| efried | cdent ^ | 17:53 |
| *** fungi has joined #openstack-placement | 19:05 | |
| *** artom has quit IRC | 21:06 | |
| *** jaypipes has quit IRC | 21:44 | |
| openstackgerrit | Eric Fried proposed openstack/placement master: Extra gabbi tests for same_subtree https://review.opendev.org/669930 | 22:00 |
| *** mriedem has quit IRC | 23:08 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!