| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2019-02-19 | |||
| 08:49:01 | openstackgerrit | Artem Goncharov proposed openstack/python-openstackclient master: Add possibility to filter images using member_status https://review.openstack.org/637508 | |
| 09:05:56 | openstackgerrit | zhongshengping proposed openstack/openstackclient master: add python 3.7 unit test job https://review.openstack.org/637739 | |
| 09:05:58 | openstackgerrit | zhongshengping proposed openstack/osc-lib master: add python 3.7 unit test job https://review.openstack.org/637743 | |
| 09:06:25 | openstackgerrit | zhongshengping proposed openstack/cliff master: add python 3.7 unit test job https://review.openstack.org/637766 | |
| 09:06:58 | openstackgerrit | zhongshengping proposed openstack/python-openstackclient master: add python 3.7 unit test job https://review.openstack.org/637785 | |
| 11:08:47 | mordred | gtema: jinx. we both rechecked that patch just now :) | |
| 11:09:08 | gtema | nice | |
| 11:09:17 | gtema | double is better | |
| 11:10:38 | mordred | hope so | |
| 11:11:00 | gtema | yeah. | |
| 11:13:00 | gtema | btw mordred, I would need designate for functional tests of dns resouces. We currently have it only in `openstacksdk-functional-devstack-networking` | |
| 11:13:19 | gtema | do you think it is sufficient or should we add something else? | |
| 11:14:53 | mordred | gtema: I think it's sufficient ... as long as we have an env where the tests run | |
| 11:15:11 | gtema | oki, thks | |
| 11:46:06 | openstackgerrit | Artem Goncharov proposed openstack/python-openstackclient master: add multiple attributes to image from APIv2.7 https://review.openstack.org/637528 | |
| 12:06:14 | openstackgerrit | David Rabel proposed openstack/python-openstackclient master: Fix help message of image add project https://review.openstack.org/637828 | |
| 12:13:48 | mordred | gtema: comments on https://review.openstack.org/#/c/637538 - sorry, I wrote the yesterday but forgot to click the button) | |
| 12:14:25 | gtema | no problem. | |
| 12:14:47 | gtema | I was already nerveous about "os_hidden" and "is_hidden" ;-) | |
| 12:15:27 | gtema | I somehow think it might even strenghten confusion | |
| 12:15:33 | gtema | but ok | |
| 12:16:38 | mordred | gtema: os_hidden is just awesome | |
| 12:17:05 | gtema | mordred: what about "is_os_hidden"? | |
| 12:17:11 | gtema | lol | |
| 12:17:39 | gtema | indeed flu | |
| 12:17:55 | mordred | gtema: maybe if glance v2 hadn't gotten rid of the properties dict for user-settable properties, there would have been a decent story for glance to add new system properties without needing to prefix them with os_ | |
| 12:18:50 | gtema | mordred: yeah, I am also not happy about all that. But anyway, making it "is_hidden", right? | |
| 12:19:57 | mordred | gtema: yeah. we try to avoid such sins as os_hidden in sdk, which may or may not be smart, but then, we also may or may not be smart :) | |
| 12:20:33 | gtema | ok, "I feel it comin" :) | |
| 12:20:59 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Add image attributes from v2.7 https://review.openstack.org/637538 | |
| 12:21:24 | mordred | beatiful | |
| 12:21:51 | gtema | mordred: also started implementing "stage" and "import" for glance yesterday - also very nice things | |
| 12:22:12 | mordred | oh cool - yeah, I've been meaning to work on those | |
| 12:22:18 | gtema | I also need around 30 hours in a day to be able to work at all | |
| 12:23:06 | mordred | gtema: did I mention aligning the shade and sdk image methods (specifically, replacing the sdk methods with the ones in the shade layer)? | |
| 12:23:13 | mordred | gtema: I started on it here: https://review.openstack.org/#/c/609684/ | |
| 12:23:17 | gtema | wrt osc: as I wrote in the maillist I would be doing that. And Glance is the beginning, therefore those patches | |
| 12:23:20 | mordred | but that's in merge conflict and needs to be updated | |
| 12:23:28 | mordred | gtema: ah - awesome! | |
| 12:23:31 | gtema | oh great | |
| 12:23:36 | mordred | gtema: sorry, I missed that from the mailing list | |
| 12:23:45 | gtema | no worry | |
| 12:24:20 | gtema | probably wrong subject prefix ;) | |
| 12:25:00 | gtema | mordred, have you seen my query in maillist about "*" region for identity? | |
| 12:25:10 | mordred | no - I'll go look for that | |
| 12:25:49 | gtema | mordred: http://lists.openstack.org/pipermail/openstack-discuss/2019-February/002777.html | |
| 12:26:27 | gtema | is somehow a tricky setup in my case which makes some troubles, but there are no strict regulations from OpenStack about that | |
| 12:27:07 | mordred | yeah. I agree - we should make a solution :) | |
| 12:27:22 | gtema | yupp | |
| 12:27:52 | mordred | gtema: I think the 'easy' workaround for the short term is to just register the identity endpoints in the catalog for each region - even if it's the same endpoint | |
| 12:27:59 | mordred | but I agree completely that this is lame | |
| 12:28:14 | mordred | and we should actually support the thing you're talking about | |
| 12:28:20 | gtema | mordred: I agree this is an "easy" solution from the implementation POV | |
| 12:28:27 | mordred | in a real non-hacky way | |
| 12:28:29 | gtema | but I need to convince operations about that | |
| 12:28:36 | gtema | and this is not trivial | |
| 12:28:41 | mordred | yeah. it never is :) | |
| 12:29:09 | mordred | would an API-SIG document describing it as a 'correct' choice help? | |
| 12:29:32 | gtema | yes, I think it would be | |
| 13:43:25 | openstackgerrit | Merged openstack/openstacksdk master: Make sure we pick flavors with disk https://review.openstack.org/637618 | |
| 13:43:44 | dtantsur | \o/ | |
| 13:44:06 | dtantsur | mordred: could you check https://review.openstack.org/#/c/636601/ if you have a minute? | |
| 13:48:00 | mordred | dtantsur: lgtm! | |
| 13:48:38 | dtantsur | cool | |
| 13:56:45 | dulek | Hi folks! Can I somehow use openstacksdk to just do plain REST calls to OpenStack endpoint, so I won't worry that e.g. there's no support for Octavia tags yet? | |
| 13:57:37 | dulek | Bare requests module won't do all the keystone authentication for me, that's why I'm looking for some middle layer here. | |
| 13:57:50 | dtantsur | dulek: I think you need the keystoneauth library | |
| 13:58:08 | dtantsur | dulek: https://docs.openstack.org/keystoneauth/latest/using-sessions.html | |
| 13:58:36 | mordred | dulek: you can totally make REST calls to an openstack endpoint | |
| 13:59:14 | dtantsur | dulek: openstacksdk also provided keystoneauth Adapters for known services, e.g. connection.baremetal is an adapter for Ironic. | |
| 13:59:24 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Add image attributes from v2.7 https://review.openstack.org/637538 | |
| 13:59:43 | mordred | dulek: each service type has a corresponding adapter on the Connection object - so you should be able to do "conn.load_balancer.get('/something/in/octavia') and it'll return a requests.Response | |
| 13:59:52 | dulek | dtantsur: Ooooh, this would probably be perfect… | |
| 14:00:09 | dulek | dtantsur: So I won't need to deal with endpoints names and stuff. | |
| 14:00:13 | mordred | yah | |
| 14:00:18 | dtantsur | yep, see mordred's example | |
| 14:00:21 | dulek | mordred, dtantsur: Awesome, thank you guys! | |
| 14:00:26 | mordred | \o/ | |
| 14:00:36 | mordred | ok. I've been helpful for today - I can go back to bed now | |
| 14:00:52 | dtantsur | ++ about the right time | |
| 14:06:07 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: WIP Move image methods to sdk image proxy https://review.openstack.org/609684 | |
| 14:12:04 | gtema | mordred: just rebased your change 609684 on top of master leaving all other related changes off, since there were lots of conflicts | |
| 14:44:24 | mordred | gtema: sweet. obviously the code needs some cleanup to start using Image resource objects internally - but I think if we can get that passing tests landing it in that state should let us Resource-ify it in smaller patches | |
| 14:46:55 | gtema | great | |
| 14:47:42 | gtema | mordred: let's see what the tests show now. Do you want to continue on it, or should I take it? | |
| 14:56:30 | johnsom | dulek: fyi, Octavia tags support has merged. Still waiting on the rest of those patches however. | |
| 15:02:23 | dulek | johnsom: Oh, I see, it's even released. :) | |
| 15:19:27 | Shrews | dtantsur: +A'd https://review.openstack.org/636594 but left a nit there you might want to see | |
| 15:22:51 | openstackgerrit | Vishakha Agarwal proposed openstack/python-openstackclient master: Cleanup: Remove deprecated role list filters https://review.openstack.org/637166 | |
| 15:24:13 | dtantsur | Shrews: thanks! I'll check that and also the place I copy-pasted it from :) | |
| 15:42:16 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Add image.task resource https://review.openstack.org/637913 | |
| 16:13:54 | openstackgerrit | Merged openstack/python-openstackclient master: Fix help message of image add project https://review.openstack.org/637828 | |
| 16:21:32 | mordred | gtema: if you haven't started yet - I'm going to take a quick stab at the failures on that WIP image reorg patch | |
| 16:22:48 | mordred | Shrews, dtantsur: https://review.openstack.org/#/c/637563/ if you have a sec ... odyssey4me was having a hard time debugging a failure yesterday in nodepool, and it's a failure mode that is always confusing to users when it happens | |
| 16:23:42 | mordred | Shrews, dtantsur: it's a behavior change, but I **seriously** doubt it's a behavior change anyone is depending on because I think I might be the only person to have that whole behavior stack in my head :) | |
| 16:24:53 | mordred | gtema: ^^ you might also want to look at that one | |
| 16:25:22 | dtantsur | mordred: any unit tests to update? | |
| 16:25:46 | mordred | dtantsur: no - it worked fine (although that maybe indicates there is a missing unittest) | |
| 16:25:55 | dtantsur | exactly ;) | |
| 16:27:27 | mordred | the only use case I could see this making harder is if a user has a cloud that only has an unknown verion of a known service - but even in that case they can set an endpoint_override and a version override and it'll work | |