Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-06-23
16:24:58 openstackgerrit Jay Pipes proposed openstack/nova master: placement: scheduler uses allocation candidates https://review.openstack.org/476631
16:24:59 openstackgerrit Jay Pipes proposed openstack/nova master: Add uuid to scheduler's HostState object https://review.openstack.org/476630
16:24:59 openstackgerrit Jay Pipes proposed openstack/nova master: placement: support GET /allocation_candidates https://review.openstack.org/475448
16:25:00 openstackgerrit Jay Pipes proposed openstack/nova master: WIP - placement: claim resources in scheduler https://review.openstack.org/476632
16:27:08 leakypipes efried, edleafe: addressed your comments (thanks!) from the reviews above.
16:27:22 efried ack
16:27:32 efried leakypipes FYI, still working my way through the series.
16:27:56 leakypipes efried: you definitely need a casual Friday nick.
16:28:09 leakypipes and with a name like Eric Fried, you have many options.
16:28:13 efried leakypipes My name anagrams to fried_rice
16:28:20 leakypipes :)
16:28:23 leakypipes perfect.
16:32:09 dtp lol
16:33:59 openstackgerrit Maciej Kucia proposed openstack/nova master: WIP Introduced class PciDeviceAddress https://review.openstack.org/476972
16:34:00 openstackgerrit Maciej Kucia proposed openstack/nova master: WIP Introduced PciDeviceAddressPattern class https://review.openstack.org/476973
16:36:57 cfriesen_ I've got a proposed change (https://review.openstack.org/#/c/461456) to treat "hw:cpu_realtime_mask" in the flavor extra-specs as a bitmask rather than an implied exclusion mask. If anyone is opinionated on this please feel free to review it. (Personally I think the implicit "all cpus are realtime" is confusing and illogical.)
16:40:17 mriedem edleafe: DictOfStringsField does not validate the value is not null
16:40:18 mriedem http://paste.openstack.org/show/613526/
16:41:04 openstackgerrit Ildiko Vancsa proposed openstack/nova master: WIP/POC Implement new attach Cinder flow https://review.openstack.org/330285
16:43:47 dtp mriedem i see your -2 on https://review.openstack.org/#/c/464280/; should i still look it over?
16:44:59 mriedem dtp: i think you already looked at the os-services one once
16:45:11 mriedem dtp: the -2 is procedural
16:45:33 mriedem once i'm done with the os-hypervisors change on top and it's generally ok with reviewers, then we can merge them in order or squash them
16:45:50 mriedem i just didn't want another microversion slipping in between
16:46:03 dtp sounds good. i'll look at the other two!
16:49:10 cfriesen_ mriedem: speaking of microversions...https://review.openstack.org/#/c/435141 is ready I think.
16:51:42 mriedem dtp: the hypervisors one just needs the unit tests done, which i'll do whilst flying today
16:52:00 mriedem cfriesen_: ok
16:52:12 dtp ok
16:58:04 mriedem flight is delayed 2 hours for maintenance and the connection was 2 hours, so will probably miss the connection to china, so maybe i'll just be hanging out in an airport reviewing code for the next 24 hours
16:59:23 mriedem artom: do you want to just fix this quick or patch it on top? https://review.openstack.org/#/c/415622/55/nova/exception.py
17:00:07 superdan mriedem: ouch
17:01:15 mriedem those 2 spaces in the middle of the error message? i know.
17:01:31 mriedem maybe i'll catch a ride down to portland for the evening
17:03:17 superdan mriedem: where are you?
17:05:38 artom mriedem, the space? Bah, might as well fix it
17:05:56 mriedem artom: just fix it quick and i'll re-+2
17:05:59 mriedem superdan: in MSP right now
17:06:05 mriedem was supposed to leave about an hour ago
17:06:13 superdan ah
17:06:13 mriedem superdan: connecting in seattle
17:06:24 mriedem had a 2 hour connection, now it's going to be like a 2 minute connection
17:06:27 superdan that's what I meant, where are you going to be
17:06:47 superdan mriedem: well, yeah if you end up in PDX we'll be glad to show you a good time :P
17:07:06 artom mriedem, done
17:07:19 artom mriedem, is Huawei bringing you in to the motherland?
17:07:24 mriedem artom: yup
17:07:36 openstackgerrit Artom Lifshitz proposed openstack/nova master: Plumbing for tagged nic attachment https://review.openstack.org/415622
17:07:39 openstackgerrit Artom Lifshitz proposed openstack/nova master: Delete nic metadata when detaching interface https://review.openstack.org/435387
17:07:48 openstackgerrit Artom Lifshitz proposed openstack/nova master: Delete disk metadata when detaching volume https://review.openstack.org/435082
17:07:50 openstackgerrit Artom Lifshitz proposed openstack/nova master: API support for tagged device attachment https://review.openstack.org/376021
17:17:02 melwitt dtp made me confused about the compute node marker thing again
17:17:40 melwitt does passing the marker mean return the one with the marker or the one after it?
17:18:01 superdan should be after it
17:18:10 melwitt okay, then that makes sense
17:18:39 melwitt ah "select = select.where(cn_tbl.c.id > marker)"
17:18:59 melwitt PHEW
17:38:52 dtp thanks melwitt for clearing that up
18:02:45 dtp are the api-ref files read by normies or are they just for nova devs?
18:13:30 cfriesen_ dtp: anyone coding to the HTTP API
18:13:44 dtp thanks
18:13:46 cfriesen_ but that could be end-users, not openstack devs
18:23:20 mriedem cfriesen_: some small issues in https://review.openstack.org/#/c/435141/ but +2
18:28:39 openstackgerrit Merged openstack/nova master: Remove code that produces warning in modern Python https://review.openstack.org/476952
18:37:16 openstackgerrit Matt Riedemann proposed openstack/python-novaclient master: Microversion 2.48: Standardization of VM diagnostics https://review.openstack.org/473228
18:42:07 cfriesen_ mriedem: for the "do the version check once", were you thinking of storing it in something like "have_embedded_flavor_info" or something like "have_version_2_47"?
18:42:55 mriedem cfriesen_: either
18:43:02 mriedem don't care as much about the name of the variable
18:44:22 cfriesen_ mriedem: as for re-using _expand_dict_key(), I tried but "flavor" doesn't support delattr/getattr/setattr
18:45:06 cfriesen_ mriedem: it's logically the same thing, but the types don't work
18:48:17 artom cfriesen_, for what it's worth, I'm a fan of indicating meaning rather than numeric version numbers
18:49:17 cfriesen_ artom: was leaning that way myself
18:56:42 openstackgerrit Ildiko Vancsa proposed openstack/nova master: WIP/POC Implement new attach Cinder flow https://review.openstack.org/330285
19:41:12 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Handle uuids in os-hypervisors API https://review.openstack.org/471899
19:43:48 openstackgerrit Merged openstack/nova master: Fix misuse of assertIsNone https://review.openstack.org/476736
20:09:40 mriedem mgagne: it's close https://review.openstack.org/#/c/454322/
20:09:58 mriedem mostly easy things to address
20:11:49 openstackgerrit Jay Pipes proposed openstack/nova master: placement: scheduler uses allocation candidates https://review.openstack.org/476631
20:11:49 openstackgerrit Jay Pipes proposed openstack/nova master: Add uuid to scheduler's HostState object https://review.openstack.org/476630
20:11:50 openstackgerrit Jay Pipes proposed openstack/nova master: placement: support GET /allocation_candidates https://review.openstack.org/475448
20:11:50 openstackgerrit Jay Pipes proposed openstack/nova master: WIP - placement: claim resources in scheduler https://review.openstack.org/476632
20:16:16 fried_rice leakypipes https://review.openstack.org/#/c/475448/9..11/nova/api/openstack/placement/handlers/allocation_candidate.py@160 -- Is there something preventing the request containing `resources=` (i.e. an empty value)?
20:16:28 fried_rice leakypipes If not, ^ will still raise ValueError
20:18:01 leakypipes fried_rice: the JSONSChema validation.
20:18:16 fried_rice leakypipes Which part of that ensures it's nonempty?
20:19:19 fried_rice leakypipes I would have expected "required": [ "resources" ] would just ensure the `resources` key is present.
20:19:54 fried_rice Because an empty value could be valid.
20:22:35 fried_rice leakypipes Just tried it out locally: `jsonschema.validate({'resources': ''}, schema, format_checker=jsonschema.FormatChecker())` passes
20:31:37 leakypipes fried_rice: then we'll have to go and change all the schema validators in all the handlers in placement...
20:32:03 fried_rice leakypipes Or just catch ValueError and convert it to HttpBadRequest
20:32:13 leakypipes fried_rice: which I can try and do this weekend. but for now, I'm tired. :)
20:35:12 leakypipes fried_rice: hold up, pushing a new patch to add that check to normalize_resources_qs_param
20:35:37 fried_rice leakypipes ight.
20:36:34 fried_rice leakypipes Yuh, that'd do it. Since that guy's clearly allowed to raise HttpBadRequest
20:41:01 mriedem melwitt: do you get a 406 or a 405 from placement if requesting a microversion that doesn't exist?
20:41:19 mriedem your consumers/allocations code is checking for a 406 response, but i thought it would be a 405
20:41:33 openstackgerrit Jay Pipes proposed openstack/nova master: move resources_from_request_spec() to utils https://review.openstack.org/476629
20:41:34 openstackgerrit Jay Pipes proposed openstack/nova master: placement: scheduler uses allocation candidates https://review.openstack.org/476631
20:41:34 openstackgerrit Jay Pipes proposed openstack/nova master: Add uuid to scheduler's HostState object https://review.openstack.org/476630
20:41:35 openstackgerrit Jay Pipes proposed openstack/nova master: placement: support GET /allocation_candidates https://review.openstack.org/475448

Earlier   Later