| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-06-20 | |||
| 12:34:05 | openstackgerrit | Merged openstack-infra/shade master: add release notes to README.rst https://review.openstack.org/576756 | |
| 13:05:08 | mnaser | mordred: do you wanna re-workflow this to avoid the trip through check/gate? https://review.openstack.org/#/c/574747/ | |
| 13:14:48 | mordred | mnaser: yup | |
| 13:14:50 | mordred | mnaser: done | |
| 13:17:43 | mnaser | mordred: cool thanks | |
| 13:27:44 | openstackgerrit | Telles Mota Vidal Nóbrega proposed openstack/python-openstackclient master: Adding api_version to FakeApp https://review.openstack.org/576833 | |
| 13:47:05 | openstackgerrit | Merged openstack/openstacksdk master: Support port binding extended attributes for querying port https://review.openstack.org/573645 | |
| 14:55:42 | cdent | edleafe, elmiko, dtantsur : question about error codes (of the "code": "placement.undefined_code" sort defined in the errors guideline) | |
| 14:56:20 | cdent | this one : http://specs.openstack.org/openstack/api-wg/guidelines/errors.html | |
| 14:57:27 | cdent | there are three different kinds of 404 one could get: /foobar is a url that never existed and never will, /reshaper is a url that exists but is a 404 at the current microversion, /resource_provider/some_uuid is that there is no resource provider with that uuid | |
| 14:57:45 | cdent | 404 is intentionally ambiguous about these differences | |
| 14:57:59 | cdent | should the error code clarify or mantain the ambiguity? | |
| 15:08:39 | mordred | cdent: clarify | |
| 15:09:12 | cdent | i'm leaning that way too | |
| 15:09:20 | mordred | cdent: one of my biggest complaits about early openstack deployments was that people would shut api calls off with policy and then have them return 404 as if the API wasn't a publically documented thing anyway | |
| 15:09:28 | mordred | but while this isn't related to that | |
| 15:09:55 | mordred | the failure case was similar - which is that as a user I wasn't sure if I'd made a mistake in constructing my url or if a resource didn't exist | |
| 15:10:11 | mordred | ( did I do /server/foo when I should have done /servers/foo ? ) | |
| 15:11:04 | mordred | SO - if there was a clarifying code in the response body "route /server is unknown" vs "server foo doesn't exist" you know | |
| 15:11:23 | dtantsur | ++ for clarifying as much as possible | |
| 15:11:50 | mordred | I think it might also make sense to indicate that it's in-game to update error responses for older microversions | |
| 15:12:00 | mordred | and that that's not an API break ... | |
| 15:12:28 | mordred | (so that you can potentially do the thing you mentioned - this route isn't valid in this microversion) | |
| 15:13:39 | cdent | i now have a todo item | |
| 15:13:45 | cdent | small but important progress or something | |
| 15:36:08 | dhellmann | dtroyer : I sat down a couple of weeks ago to look at ripping cmd2 out of cliff, but got bogged down in the amount we seem to rely on its differences in interactive mode. | |
| 15:36:10 | dhellmann | :-/ | |
| 15:38:48 | dtroyer | interactive mode is one of those things that if you use it you'll miss it but I don't think the majority of OSC users use it | |
| 15:39:12 | dtroyer | I did start converting DevStack to use it more a while back to try and sidestep the startup time issue | |
| 15:39:25 | dtroyer | piping in a list of commands | |
| 15:39:45 | gtema | I am a total fan of interactive mode | |
| 15:39:50 | dtroyer | if we decide to rip it out, OSC4 would be the time | |
| 15:40:05 | dtroyer | ya, I really like it, which is why it exists | |
| 15:40:38 | dtroyer | really, the issue comes down to the 2 or 3 times a year cmd2 causes us to have to jump and if the effort of changing is worth it | |
| 15:41:29 | openstackgerrit | Merged openstack/cliff master: support cmd2 0.9.1 in interactive mode https://review.openstack.org/571524 | |
| 15:48:59 | edleafe | cdent: sorry, today is meeting hell for me. | |
| 15:49:29 | cdent | edleafe: I forgive you | |
| 15:49:40 | edleafe | cdent: I'm all for clarity in error codes, so if it is possible to help the user with more information, +1 from me | |
| 16:04:07 | openstackgerrit | Ken Giusti proposed openstack/cliff master: Add test case for interactive help https://review.openstack.org/576900 | |
| 16:31:33 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Adds Senlin support to openstacksdk https://review.openstack.org/576650 | |
| 16:32:31 | mordred | dtantsur: ^^ I just went in an cleaned up the things you pointed out (except for docstrings) | |
| 16:33:15 | dtantsur | cool! | |
| 16:34:12 | mordred | dtantsur: thanks for that review, btw - my eyeballs had glazed over and missed those :) | |
| 16:34:20 | dtantsur | np :) | |
| 16:57:47 | elmiko | cdent: little late to the party, but i agree with the clarify position | |
| 21:21:54 | flwang | mordred: hello, I have some questions about Magnum support in Shade | |
| 21:23:16 | flwang | mordred: i can see shade is still using old "baymodels", so is there any plan to support the new cluster templates? I would like to contribute but just wanna figure out the deprecation process, thanks | |
| 22:55:39 | mordred | flwang: it shuld also have calls that look like get_cluster_template api calls - but yes, they call the older baymodel endpoint | |
| 22:56:31 | mordred | flwang: I'd love help making the magnum support better though | |
| 22:57:04 | mordred | flwang: is there a way to detect of baymodel or cluster_template should be used? | |
| 22:59:00 | mnaser | is alan bishop on irc? | |
| 22:59:11 | mordred | his name is robert paulson | |
| 22:59:34 | mnaser | i want to give him a hi5 for tracking this down.. it took me ages and i couldnt find it.. https://review.openstack.org/#/c/552814/ | |
| 23:01:22 | mnaser | also i've had a long day and i just caught that on mordred | |
| 23:03:33 | mordred | mnaser: :) | |
| #openstack-sdks - 2018-06-21 | |||
| 00:00:46 | flwang | mordred: i don't there is a good way to do auto detect | |
| 00:01:35 | flwang | mordred: so i'd like to figure out what's the correct way to do the migration in shade | |
| 00:02:09 | mordred | flwang: nod. I'm EOD today - lemme think about it and leave you a note tomorrow or so | |
| 00:02:53 | flwang | mordred: no problem. thanks | |
| 03:26:39 | openstackgerrit | John Haan proposed openstack/python-openstackclient master: Implement image tag into OSC https://review.openstack.org/576804 | |
| 08:54:30 | openstackgerrit | Dongcan Ye proposed openstack/python-openstackclient master: Fix volume type functional tests https://review.openstack.org/575448 | |
| 09:59:20 | openstackgerrit | Chris Dent proposed openstack/api-wg master: Expand error code document to expect clarity https://review.openstack.org/577118 | |
| 11:42:24 | mordred | flwang: actually - clustertemplate got added to magnum a LONG time ago didn't it? | |
| 11:43:14 | mordred | flwang: I think what I'd suggest is something similar to what we do for images - where we support nova image api if the glance one isn't there ... | |
| 11:44:22 | mordred | flwang: and that's to update the code to call the clustertemplate endpoints - but if we get a 404 back, try calling the baymodel endpoint instead. it's an extra api call cost for anybody still using a magnum old enough to not have clustertemplates - but that should be almost nobody | |
| 11:47:08 | mordred | flwang: then, as an optimization, we can make a flag "self._has_clustertemplates" or something that we can set to true if we ever make a successful call to a clustertemplates endpoint - and if it's true we can skip calls to baymodels - because we've already discovered that the cloud is new enough to have clustertemplates endpoints | |
| 11:47:39 | tellesnobrega | good morning, I sent a patch yesterday to add a variable to FakeApp but it is failing on the gate with some issues volume, the patch is this one https://review.openstack.org/#/c/576833/ | |
| 11:48:20 | tellesnobrega | I'm quite sure the issue is not related to what I did, but I'm not very familiar with openstackclient so, if someone could help me out see the problem I would really appreciate | |
| 11:49:24 | mordred | tellesnobrega: dtroyer has been working on fixing the volumetype tests | |
| 11:49:39 | mordred | tellesnobrega: here: https://review.openstack.org/#/c/575448/ | |
| 11:49:56 | tellesnobrega | mordred, awesome, thanks. just wanted to be sure I didn't break the whole thing | |
| 11:50:08 | mordred | nope - not you :) | |
| 11:50:24 | tellesnobrega | cool | |
| 11:50:39 | tellesnobrega | not this time anyway, you never know | |
| 11:50:41 | tellesnobrega | :) | |
| 12:07:53 | mordred | hehe | |
| 12:24:06 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Adds Senlin support to openstacksdk https://review.openstack.org/576650 | |
| 12:26:16 | openstackgerrit | Jens Harbott (frickler) proposed openstack/python-openstackclient master: Fix volume type functional tests https://review.openstack.org/575448 | |
| 12:27:23 | frickler | dtroyer: ^^ one step further, but there is still seeing something fishy with the v3 volume type stuff. the c-api logs errors about "'all_tenants' filter key is not valid." but I cannot find where this comes from | |
| 12:28:05 | frickler | reading even | |
| 12:29:36 | mordred | frickler: I find it better to never think about or read anything I type | |
| 12:30:58 | frickler | mordred: words of wisdom indeed | |
| 12:31:50 | frickler | mnaser: I found some log in #os-cinder where you seemed to have a similar issue with volume types, did you find a solution for that? | |
| 12:41:15 | frickler | mnaser: dtroyer: I found a reference to the patch in python-cinderclient that breaks us and created https://bugs.launchpad.net/python-cinderclient/+bug/1778055 , when that is fixed we should be good again together with the patch above. | |
| 12:41:16 | openstack | Launchpad bug 1778055 in python-cinderclient "Change in search behaviour for v3 volume type breaks OSC" [Undecided,New] | |
| 12:41:37 | frickler | I think we should disable volume type v3 tests until then | |
| 12:57:55 | mnaser | frickler: Thank you | |
| 13:01:07 | rcarrillocruz | hey folks | |
| 13:01:56 | frickler | mnaser: I also just confirmed that reverting the patch in question would fix the issue for me locally, but I'll give cinder folks a bit to come up with a better solution | |
| 13:01:58 | rcarrillocruz | does https://docs.openstack.org/keystone/queens/user/application_credentials.html work since queens onwards? And is there a need to have some specific setting on keystone config to use that from end user perspective or providing Queens/ |
|
| 13:02:08 | rcarrillocruz | mnaser, mordred ^ | |
| 13:02:22 | rcarrillocruz | you deal with quite a bit of clouds :-) | |
| 13:03:05 | mnaser | frickler: yeah it quite most definetly is the cause. Let’s see what happens | |
| 13:03:27 | mnaser | rcarrillocruz: i think it’s just a Queens feature | |
| 13:03:51 | rcarrillocruz | mnaser: where's vexxhost version at? | |
| 13:04:05 | mnaser | Queens :) | |
| 13:04:11 | rcarrillocruz | i'm golden :-) | |
| 13:04:15 | rcarrillocruz | thanks | |
| 13:04:36 | mnaser | Let me know how it goes. I think cmurphy might be interested to see your experience too :) | |
| 13:05:39 | rcarrillocruz | ++ | |