| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-07-31 | |||
| 14:14:02 | mordred | cool - thanks! I *think* it might be a list of dicts with a message key - since we've got sdk code to handle that case | |
| 14:16:40 | cdent | mordred: looks like a dict keys on the fault name: https://github.com/openstack/nova/blob/03ccb1f9333bcfbdb2caf8a8c6e98622a63088cc/nova/api/openstack/wsgi.py#L1080-L1083 | |
| 14:17:25 | mordred | cdent: ah - cool. that would explain the code in sdk currently | |
| 14:17:30 | mordred | I will add a comment | |
| 14:22:53 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Add support for neutron error messages https://review.openstack.org/587483 | |
| 14:34:44 | mordred | cdent: ^^ just in case you are curious. | |
| 14:42:14 | cdent | I'm sure we made this http://specs.openstack.org/openstack/api-wg/guidelines/errors.html for some reason but I can't remember why | |
| 15:01:37 | openstackgerrit | Merged openstack/openstacksdk master: Support for microversions in baremetal resources https://review.openstack.org/583532 | |
| 15:01:39 | openstackgerrit | Merged openstack/openstacksdk master: Add a simple baremetal functional job https://review.openstack.org/586574 | |
| 20:13:11 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: [WIP] Fix broken gate jobs. https://review.openstack.org/587005 | |
| 20:19:13 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: [WIP] Fix broken gate jobs. https://review.openstack.org/587005 | |
| 20:23:32 | dtroyer | mordred: ^^^ should get us really close… it needs a way to detect Image API v1 yet | |
| 20:24:05 | mordred | dtroyer: like - needs to be able to detect whether the cloud has v1? | |
| 20:24:41 | dtroyer | yes | |
| 20:24:55 | mordred | https://review.openstack.org/#/c/585029/ <-- | |
| 20:25:13 | mordred | dtroyer: that is what you need - unfortunately I did not manage to get it in to keystoneauth before the feature freeze | |
| 20:25:29 | dtroyer | I've hard-coded it 'not there' in that rev just to get this through… ah, nice | |
| 20:25:52 | mordred | dtroyer: yah - if you're ok with the test suite being a little wonk until we can get another ksa release cut - there's a path to it | |
| 20:26:09 | dtroyer | I'll think about this harder after the 30,000 foot fatigue is less | |
| 20:26:28 | dtroyer | I'm fine, this is better than ripping it all out | |
| 20:26:33 | mordred | dtroyer: alternately, the code in the versions show command that does it client-side would also work - it'll just be more expensive than would normally be pleasant | |
| 20:26:44 | mordred | cool | |
| 20:27:07 | dtroyer | hmmm… yeah, I was thinking something like that, it's a one-time deal so we can live with it | |
| 20:27:49 | mordred | ++ | |
| 20:28:07 | mordred | and it should all be really cheap local calls in functional tests - not quite like doing the same thing on an actual public cloud :) | |
| 20:58:37 | samueldmq | hi. I am not getting why we need ansible tests for the upper abstraction level/shade | |
| 20:58:49 | samueldmq | as opposed to having just the unit and functional tests... | |
| 20:59:24 | samueldmq | seems like those ansible tests should belong somewhere in the shade plugin to ansible (or something like that)? | |
| 21:01:17 | mordred | samueldmq: yes - they should indeed be there | |
| 21:01:38 | mordred | samueldmq: they were in the shade repo for hysterical raisins | |
| 21:01:59 | mordred | samueldmq: mostly that we didn't have a way to run them if we put them into the ansible repo | |
| 21:02:18 | mordred | samueldmq: but now that we can (and do) trigger jobs in Infra on PRs to ansible ... | |
| 21:02:34 | mordred | samueldmq: we should TOTALLY move them to ansible/ansible and get them integrated with ansible-test | |
| 21:02:53 | samueldmq | kk should be up to them to run the tests? | |
| 21:03:04 | samueldmq | or to us on our infra (seems to be what you're suggesting) | |
| 21:03:25 | samueldmq | mordred: ^ | |
| 21:05:23 | samueldmq | ahh I got it ... we can add a check on sdks to run ansible tests against proposed shade changes | |
| 21:05:32 | samueldmq | just to make sure we're not breaking'em on our side | |
| 21:06:01 | mordred | that's right | |
| 21:06:35 | mordred | we actually already run the shade-ansible test job on ansible/ansible PRs that touch openstack bits | |
| 21:06:42 | mordred | as well as on both shade and openstacksdk | |
| 21:07:25 | samueldmq | ok it's just a matter of putting things in their right place | |
| 21:07:53 | samueldmq | for running against arbitrary cloud ... do you think I should work on shade/functional tests side only? | |
| 21:08:20 | samueldmq | shade should be able to run its tests against arbitrary clouds, regardless being useful to ansible | |
| 21:08:28 | mordred | yes - I agree | |
| 21:08:38 | mordred | (although do it in openstacksdk and not shade- but yes) | |
| 21:08:56 | mordred | we also have WAY more coverage in the openstacksdk functional tests than we do in the ansible tests | |
| 21:09:00 | mordred | so we get a much nicer win | |
| 21:09:12 | samueldmq | ++ | |
| 21:09:39 | samueldmq | yes I meant openstacksdk ... just not got used to the change yet (since I had disapeared by the time of the merge) :-) | |
| 21:10:02 | samueldmq | didnt get* | |
| 21:10:23 | samueldmq | kk I'll take a look there .. but I'm a bit worried about the policy bits... | |
| 21:10:39 | samueldmq | since different users will have different permissions on different clouds | |
| 21:11:35 | samueldmq | that's a bit different than running just on devstack (opinionated). may require some work but should not be terrible' | |
| 21:16:55 | openstackgerrit | Doug Hellmann proposed openstack/openstacksdk master: fix 2 typos in documentation https://review.openstack.org/587624 | |
| 21:28:19 | mordred | samueldmq: yes - I totally agree (and same here - I still say shade a lot) | |
| #openstack-sdks - 2018-08-01 | |||
| 01:07:32 | openstackgerrit | Fan Zhang proposed openstack/python-openstackclient master: [WIP] Fix broken gate jobs. https://review.openstack.org/587005 | |
| 01:12:13 | fanzhang | mordred dtroyer ^ there is still 3 failed tests, and I may not have time to work on it today. BTW, added co-authored-by to commit messages. Thanks for helping on it :-) | |
| 01:14:30 | dtroyer | fanzhang: sorry about my bad typing (Falsex! ha!)… the remaining failures do appear to be the cinderclient issue being addressed in https://review.openstack.org/#/c/587601/ | |
| 01:15:04 | dtroyer | Thanks for working on this, we'll keep it going one way or another | |
| 01:16:41 | fanzhang | dtroyer yeh, just noticed email of Sean. Np. ;-) | |
| 09:14:56 | openstackgerrit | tianhui proposed openstack/python-openstackclient master: Compute: Add user id support for keypair https://review.openstack.org/576507 | |
| 09:26:13 | openstackgerrit | tianhui proposed openstack/python-openstackclient master: Compute:Support keypair set key-type https://review.openstack.org/577768 | |
| 09:27:52 | openstackgerrit | yanpuqing proposed openstack/openstacksdk master: Add an error message when router add subnet https://review.openstack.org/586996 | |
| 14:43:33 | usr2033 | hi, are all sdks interact with services by rest api? are there any way to direct interaction with services? | |
| 14:56:40 | mhen | usr2033, from the SDK side only REST is available as far as I am aware. A more direct communication would be via the RPC channels but this is only for internal communication between a service's components and is not a public API for generic clients. | |
| 15:02:01 | mordred | yes, that's correct | |
| 15:02:42 | usr2033 | mhen: i tried shade. only 1 of 6 try on get_project works | |
| 15:03:09 | usr2033 | but curl, python.request works well | |
| 15:03:59 | mhen | usr2033, I'm not familiar with shade sadly, I only use the openstackclient | |
| 15:04:46 | usr2033 | i got unable to find resource, project doesnot exits | |
| 15:05:34 | usr2033 | hımm. i will try it. one sdk rules all was attractive but not working for now. thanks mhen | |
| 15:06:10 | mhen | openstackclient is using the SDK most of the time as well | |
| 15:07:20 | usr2033 | openstackclient for example nova's client uses openstacksdk? am i understand right? | |
| 15:08:56 | dtroyer | OSC only uses the SDK for network commands today | |
| 15:09:15 | dtroyer | once SDK 1.0 is out we will begin converting, starting with Glance and Cinder | |
| 15:12:13 | mhen | okay that's not "most of the time" ... thought it was more than network by now | |
| 15:13:24 | mordred | all of the ansible modules use sdk too though - I expect sdk/shade to work | |
| 15:13:47 | mordred | usr2033: it is possible that the user you are logging in with does not have access to get_project on the project you are looking for? | |
| 15:14:07 | mhen | usr2033, to correct my statement a bit: openstackclient is using the client libraries (e.g. python-novaclient, python-cinderclient, etc.) internally except for network, where it already uses the SDK. The SDK is essentially a unififed library to replace the individual client libraries in the future. | |
| 15:24:05 | mordred | yes. and I expect that any methods in sdk should work with any openstack cloud - so if they do not, please let us know so we can fix it | |
| 21:08:24 | openstackgerrit | Merged openstack/python-openstackclient master: Update reno for stable/rocky https://review.openstack.org/586006 | |
| 21:10:26 | openstackgerrit | Merged openstack/osc-lib master: Update reno for stable/rocky https://review.openstack.org/586112 | |
| #openstack-sdks - 2018-08-02 | |||
| 05:54:45 | usr2033 | mordred: i got project data on 1 of my 7 calls. | |
| 08:05:01 | openstackgerrit | Lajos Katona proposed openstack/python-openstackclient master: Detailed help message for QoS max-burst-kbps value https://review.openstack.org/588168 | |
| 12:37:46 | samueldmq | morning | |
| 12:37:58 | samueldmq | does anybody know if I can do LIBS_FROM_GIT=openstacksdk | |
| 12:38:00 | samueldmq | in devstack? | |
| 12:42:56 | dims | samueldmq : yes. that should work | |
| 12:49:54 | samueldmq | dims: nice, thanks | |
| 16:34:32 | imacdonn | mordred: around? | |
| 16:35:44 | mordred | imacdonn: yes - although I'm about to do an ops-related thing, so I might not be very responsive for a bit | |
| 16:36:26 | imacdonn | mordred: ok... will throw this out anyway ... maybe someone else has input | |
| 16:37:42 | imacdonn | trying to use new openstacksdk via ansible to create a volume, and it's coughing up "keystoneauth1.exceptions.catalog.EndpointNotFound: public endpoint for block-storage service not found" | |
| 16:37:45 | imacdonn | seems to be due to https://review.openstack.org/#/c/564494/ | |
| 16:38:03 | imacdonn | the release note for that says "There should be no behavior differences" .......... ;) | |
| 16:38:46 | mordred | there should be ;) | |
| 16:39:06 | imacdonn | ahhh, the old "should" "_ | |
| 16:39:07 | mordred | you have latest openstacksdk and latest keystoneauth installed? | |
| 16:39:08 | imacdonn | :) | |