| *** jamielennox is now known as jamielennox|away | 00:20 | |
| *** jamielennox|away is now known as jamielennox | 00:27 | |
| *** gouthamr has quit IRC | 02:17 | |
| mordred | ianw: around? https://review.openstack.org/#/c/443251 and https://review.openstack.org/#/c/444331 (the second addresses your concerns from the first) | 05:25 |
|---|---|---|
| ianw | mordred: yeah, i think they're ok ... complexity sort of increased from the original :) | 05:36 |
| mordred | ianw: yah. it did a bit didn't it :) | 05:46 |
| openstackgerrit | Merged openstack-infra/shade master: Convert test_services to requests_mock https://review.openstack.org/444475 | 06:09 |
| openstackgerrit | Merged openstack-infra/shade master: Put fnmatch code back, but safely this time https://review.openstack.org/443251 | 06:24 |
| openstackgerrit | Merged openstack-infra/shade master: Only do fnmatch compilation and logging once per loop https://review.openstack.org/444331 | 06:24 |
| openstackgerrit | Merged openstack-infra/shade master: Raise a more specific exception on nova 400 errors https://review.openstack.org/441205 | 06:24 |
| *** openstackgerrit has quit IRC | 08:18 | |
| *** yfried has joined #openstack-shade | 08:37 | |
| *** cdent has joined #openstack-shade | 11:01 | |
| *** ioggstream has joined #openstack-shade | 12:03 | |
| *** ioggstream has quit IRC | 12:13 | |
| *** DuncanT has joined #openstack-shade | 12:26 | |
| *** gouthamr has joined #openstack-shade | 13:47 | |
| *** openstackgerrit has joined #openstack-shade | 14:21 | |
| openstackgerrit | Jordan Pittier proposed openstack-infra/shade master: attach_volume should always return a vol attachment. https://review.openstack.org/434964 | 14:21 |
| *** dtroyer_zz has quit IRC | 14:33 | |
| *** dtroyer has joined #openstack-shade | 14:33 | |
| *** ioggstream has joined #openstack-shade | 15:40 | |
| *** ioggstream has quit IRC | 15:52 | |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade master: change test_endpoints to use requests mock https://review.openstack.org/444541 | 15:53 |
| *** yfried has quit IRC | 15:56 | |
| *** yfried has joined #openstack-shade | 15:56 | |
| *** yfried has quit IRC | 16:05 | |
| *** thingee has quit IRC | 16:45 | |
| notmorgan | uh... | 16:51 |
| notmorgan | mordred: so... i have what looks to be a mismatch in keystone docs and shade (checking keystone server) | 16:52 |
| notmorgan | but i think we have some doc != behavior *sigh* | 16:52 |
| notmorgan | yep, looks like shade was actually incorrect here | 16:52 |
| mordred | yay! | 16:53 |
| notmorgan | you could never create a endpoint | 16:53 |
| notmorgan | https://www.irccloud.com/pastebin/IAUEjoaM/ | 16:53 |
| notmorgan | note "region_id" vs "region" | 16:53 |
| notmorgan | something is wonky | 16:53 |
| notmorgan | but i'm digging into keystone | 16:53 |
| notmorgan | maybe we just really suck at making requests different from api results (oh god, please no) | 16:53 |
| notmorgan | hmm yeah looks like region_id is correct in all cases | 16:54 |
| notmorgan | not "region" | 16:54 |
| notmorgan | mordred: so... | 17:04 |
| notmorgan | mordred: since this never could have worked in shade | 17:04 |
| notmorgan | is it ok to change the signatures to use 'region' | 17:04 |
| notmorgan | no one could have used it without it raising exceptions/breaking anyway | 17:04 |
| mordred | notmorgan: we have functional tests that test that function though | 17:05 |
| mordred | (not arguing that you're not right - just wondering how it worked against devstacka) | 17:06 |
| mordred | also - what does region_id mean? that makes it seem like there is also a region_name somewhere :) | 17:07 |
| mordred | oh wow - and we _still_ haven't landed the ansible module for endpoints | 17:09 |
| *** openstackgerrit has quit IRC | 17:18 | |
| *** gouthamr has quit IRC | 17:21 | |
| notmorgan | region_id is all we store in the db | 17:22 |
| notmorgan | region_id is the region_name | 17:22 |
| notmorgan | or whatever is in the endpoint | 17:22 |
| notmorgan | region doesn't have a name attribute | 17:22 |
| mordred | yah. I mean - it's called region_name in the client libs on the consume-side | 17:23 |
| mordred | sigh | 17:23 |
| mordred | notmorgan: region is what keystoneclient endpoint.update takes as an arg | 17:25 |
| mordred | def create(self, service, url, interface=None, region=None, enabled=True, | 17:25 |
| mordred | **kwargs): | 17:25 |
| mordred | when we switch that to call rest instead, we'll want to change it to send region_id in the payload instead of region in the kwargs | 17:26 |
| notmorgan | yeah. it's... ugh | 17:26 |
| notmorgan | damn it | 17:26 |
| notmorgan | keystoneclient is bad and should feel bad | 17:26 |
| mordred | :) | 17:26 |
| notmorgan | god this is just terrible | 17:27 |
| notmorgan | because we now have something that doesn't even remotely mirror the API | 17:27 |
| notmorgan | fwiw, it seems like ksc does weird things. | 17:27 |
| mordred | yes, this iwll be yet-another place where the shade API and the REST API do not resemble each other | 17:28 |
| notmorgan | uhm | 17:28 |
| mordred | silly me for ever having used the python-*client | 17:28 |
| notmorgan | keystoneclient is wrong here afaict | 17:28 |
| mordred | libs | 17:28 |
| mordred | notmorgan: that does not surprise me | 17:28 |
| notmorgan | this *only* works because we accidently have "extras" on endpoint | 17:28 |
| notmorgan | oh god no. No ... no no no no no no | 17:29 |
| notmorgan | endpoint['region_id'] = endpoint.pop('region') | 17:29 |
| notmorgan | in the controller | 17:29 |
| notmorgan | this means we *expect* region to be sent | 17:30 |
| notmorgan | but we respond with region_id | 17:30 |
| mordred | notmorgan: happy monday! | 17:30 |
| * notmorgan throws hands up and goes and does something else for an hour or two | 17:31 | |
| notmorgan | i have some other things to deal with, so i am going to take a break before i get really annoyed by this. | 17:31 |
| notmorgan | but in short, shade *is* wrong. but not in the way we expected | 17:31 |
| notmorgan | s/we/i | 17:32 |
| *** gouthamr has joined #openstack-shade | 17:37 | |
| -openstackstatus- NOTICE: restarting gerrit to address performance problems | 17:45 | |
| *** jamielennox is now known as jamielennox|away | 17:53 | |
| *** openstackgerrit has joined #openstack-shade | 18:37 | |
| openstackgerrit | Clark Boylan proposed openstack/os-client-config master: OVH supports qcow2 https://review.openstack.org/445042 | 18:37 |
| SamYaple | mordred: still havent landed since we still need shade changes for endpoints that i never wrote! | 19:38 |
| SamYaple | mostly because i could never figure out how to do it cleanly | 19:38 |
| SamYaple | we came up with a working solution on paper. then i went to implement it and that broke backward compat for shade and i just sorta gave up | 19:39 |
| SamYaple | landed the ansible module* -- there, context | 19:40 |
| *** jamielennox|away is now known as jamielennox | 20:37 | |
| mordred | SamYaple: yah- based on things notmorgan has said I'm starting to think our approach to endpoints was wrong from the start | 20:44 |
| *** cdent has quit IRC | 20:47 | |
| SamYaple | mordred: i could go find the logs where i said simliar things... | 20:56 |
| SamYaple | but its not on you | 20:56 |
| SamYaple | its on keystone v2 and the keystone pyhton libs | 20:56 |
| SamYaple | oh im sorry. keystone v2.0 | 20:56 |
| SamYaple | the fact is, the 2 and 3 compat is killing it | 20:56 |
| SamYaple | because 3 operates differently than 2 in a few key areas that its almost its own thing | 20:57 |
| mordred | yah - I think in hindsight we should have just made 2 calls, since it's an operator action and the operator should know if they're running v2 or v3 - but ... water under the bridge at this point | 21:04 |
| openstackgerrit | Merged openstack/os-client-config master: OVH supports qcow2 https://review.openstack.org/445042 | 21:05 |
| *** gouthamr has quit IRC | 21:06 | |
| SamYaple | mordred: i mean, that isn't a new opinion. | 21:38 |
| SamYaple | ##ansible-shade.2016-08-08.log:14:57 < mordred> well, I'm ok with us adding new features that only work on keystone v3 in operatorcloud | 21:38 |
| SamYaple | ##ansible-shade.2016-08-08.log:14:57 < mordred> because I think it's ok for operators to know what version of keystone they're running | 21:38 |
| SamYaple | the issue was trying to return v2 and v3 in the same way | 21:38 |
| SamYaple | forcing it into the same structure | 21:38 |
| SamYaple | well my issue, the one notmorgan brings up is different | 21:38 |
| *** jamielennox is now known as jamielennox|away | 22:03 | |
| *** jamielennox|away is now known as jamielennox | 22:07 | |
| *** gouthamr has joined #openstack-shade | 22:28 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!