| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-09-19 | |||
| 08:22:13 | openstackgerrit | Ian Wienand proposed openstack/openstacksdk master: Handle empty values in regions https://review.openstack.org/601485 | |
| 08:59:29 | dtantsur | AJaeger: ouch. mordred, shade CI seems to use new ansible with old shade/openstacksdk | |
| 08:59:46 | dtantsur | so it actually does not exercise shade. need to cap ansible? | |
| 09:01:07 | dtantsur | AJaeger: I'd wait for mordred, I'm not overly familiar with shade. But I suspect we may need to upper-cap ansible on all its ansible jobs. | |
| 09:18:28 | AJaeger | dtantsur: thanks for digging into it, hope mordred has an idea. | |
| 09:57:25 | mordred | AJaeger, dtantsur: I just pushed up another rev on 594393 | |
| 09:57:36 | dtantsur | wow, an early mordred | |
| 09:57:45 | dtantsur | (or is it late?) | |
| 09:57:59 | mordred | dtantsur: it's early - and for no reason I can determine :( | |
| 09:58:16 | dtantsur | happens :( | |
| 09:58:21 | mordred | but - maybe that means I'm going to be productive today | |
| 09:58:53 | dtantsur | early wake up - and productive? gotta need a few litres of coffee :) | |
| 10:03:00 | mordred | oh. yes. very much so | |
| 10:03:21 | ml | I got some issues with the CI too, maybe you can help | |
| 10:03:34 | AJaeger | mordred: if that job passes, we might need to merge the project-config change first to remove the "broken" ansible job to merge 594393. But since 594393 is the last one, this is fine... | |
| 10:03:54 | ml | want to enable neutron-fwaas for my implementation over here: https://review.openstack.org/#/c/592303/ but didn't have any success yet | |
| 10:04:06 | openstackgerrit | Merged openstack/openstacksdk master: Add support for configured NAT source variable https://review.openstack.org/602618 | |
| 10:05:31 | ml | trying to enable neutron-fwaas for the devstack checks | |
| 10:06:07 | mordred | AJaeger: ++ | |
| 10:07:40 | mordred | ml: looking | |
| 10:07:48 | ml | thank you | |
| 10:13:52 | mordred | ml: I added a couple of comments - I think we need to move those devstack/settings into the .zuul.yaml | |
| 10:16:44 | ml | thanks. I will apply those changes right now | |
| 10:22:37 | mordred | dtantsur: my god. my brain is bleeding reading the microversion doc change - how did I ever write the initial versions of that | |
| 10:23:57 | dtantsur | LOL | |
| 10:24:11 | dtantsur | mordred: I guess you were really, REALLY inspired at that moment | |
| 10:24:27 | dtantsur | it took me quite some time to wrap my head around the flow | |
| 10:24:50 | mordred | we shoudl not let me write docs, I think :) | |
| 10:25:21 | dtantsur | nobody likes writing docs, so we probably will let you anyway :) | |
| 10:28:49 | openstackgerrit | Matthias Lisin proposed openstack/openstacksdk master: Support firewall service for SDK https://review.openstack.org/592303 | |
| 10:29:48 | ml | can you quickly confirm, that it's valid: https://review.openstack.org/#/c/592303/12..13/.zuul.yaml | |
| 10:35:34 | mordred | ml: it's possible it just wants to be service_plugins: router,metering,qos,trunk,firewall_v2 - but I'm not 100% sure on that, so let's see how it goes | |
| 10:36:20 | ml | noted. | |
| 11:20:59 | openstackgerrit | Ian Wienand proposed openstack/openstacksdk master: Handle empty values in regions https://review.openstack.org/601485 | |
| 12:02:56 | openstackgerrit | Merged openstack/os-client-config master: fix bug link in readme https://review.openstack.org/597558 | |
| 12:43:54 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Run all tasks through the threadpool https://review.openstack.org/603739 | |
| 13:04:12 | AJaeger | mordred: could you see scrollback on #openstack-infra and help promoting a job, please? | |
| 13:12:43 | openstackgerrit | Merged openstack/microversion-parse master: Copy zuul job settings from project-config https://review.openstack.org/599448 | |
| 13:12:44 | openstackgerrit | Merged openstack/microversion-parse master: Add python36 jobs to microversion-parse https://review.openstack.org/599539 | |
| 14:42:45 | openstackgerrit | Merged openstack/openstacksdk master: Listing objects to return Munch objects https://review.openstack.org/596780 | |
| 14:54:02 | ml | do new implementations require the use of tenant_id or can I completely drop it in favor or project_id? | |
| 14:58:02 | mordred | ml: please use project_id - I would prefer the word tenant showed up nowhere :) | |
| 14:58:37 | ml | excellent, less work | |
| 15:03:51 | openstackgerrit | Merged openstack/openstacksdk master: baremetal: implement validate_node https://review.openstack.org/599533 | |
| 15:07:54 | openstackgerrit | Matthias Lisin proposed openstack/openstacksdk master: Support firewall service for SDK https://review.openstack.org/592303 | |
| 15:56:34 | mordred | dtantsur: ok. I think I figure out the bifrost test failure | |
| 15:56:47 | dtantsur | \o/ | |
| 15:57:07 | dtantsur | so, what's it? | |
| 15:57:33 | mordred | dtantsur: tl;dr - when noauth and endpoint are used, things are weird - so I've got some code adding a conditional for "did the user set overrides, if so just trust the user" | |
| 15:58:00 | dtantsur | mordred: how does it play with our looooong text for api-sig? | |
| 15:58:15 | dtantsur | I remember thinking really hard on a few bits around endpoint-override.. | |
| 15:58:52 | mordred | I *think* it's correct - but we should definitely walk through it and make sure | |
| 15:59:10 | mordred | also - I think I found a minor-ish bug in keystoneauth ... but maybe it's not one | |
| 15:59:42 | mordred | kmalloc, cmurphy: in keystoneauth's discovery cache, we cache http://localhost:6385/ separately from http://localhost:6385 | |
| 16:00:17 | mordred | so if someone puts in an endpoint_override with http://localhost:6385 and then the discovery document or catalog has http://localhost:6385/ - I *think* we're ignoring the cache and making a second call | |
| 16:00:21 | kmalloc | *blink* | |
| 16:00:42 | cmurphy | yay | |
| 16:02:30 | mordred | cmurphy: not me not me | |
| 16:02:30 | dtantsur | Oo | |
| 16:02:48 | kmalloc | mordred: so... Fwiw, the URL with the trailing slash is often different. With exception of the first slash | |
| 16:02:59 | kmalloc | So sounds like we need an rstrip | |
| 16:03:23 | kmalloc | mordred: it was either you or jamielennox I think. | |
| 16:03:28 | kmalloc | mordred: :P | |
| 16:03:39 | mordred | kmalloc: I'm going to blame jamielennox because he's not going to defend himself which is safer for me | |
| 16:03:43 | kmalloc | Heheheh | |
| 16:03:52 | openstackgerrit | Monty Taylor proposed openstack/openstacksdk master: Use discovery instead of config to create proxies https://review.openstack.org/600292 | |
| 16:04:53 | mordred | dtantsur, kmalloc, cmurphy: https://review.openstack.org/#/c/600292/5..6/openstack/service_description.py - wit that change, openstack.baremetal is the correct proxy object in a devstack | |
| 16:05:49 | kmalloc | My eyes are bleeding. | |
| 16:06:01 | mordred | they should be | |
| 16:06:14 | mordred | that means you still have some amount of sanity | |
| 16:06:42 | kmalloc | Oh it's ok, I lost most of my sanity in the 10000 lines for flask | |
| 16:07:09 | kmalloc | My eyes are only bleeding because of the number of files changed and the number of loc. | |
| 16:07:14 | kmalloc | Not the content of the patch. | |
| 16:08:16 | dtantsur | mordred: will it work with an unversioned endpoint? | |
| 16:08:17 | mordred | kmalloc: MOST of it is removing useless lines from tests | |
| 16:08:23 | kmalloc | mordred: exactly | |
| 16:08:29 | mordred | dtantsur: I believe so | |
| 16:09:04 | mordred | dtantsur: well, I say that ... but actually no, I think it still won't | |
| 16:09:18 | mordred | I think there is still one more "bug" preventing that from working | |
| 16:09:24 | dtantsur | \o/ | |
| 16:09:45 | mordred | which is - in most of the codepaths in keystoneauth if someone gives an endpoint_override we treat it as unassailable fact | |
| 16:09:57 | mordred | that means we don't do discovery on it | |
| 16:10:06 | mordred | since the user gave us the endpoint | |
| 16:10:11 | dtantsur | aha, and this is something that I think I changed in the API SIG patch ;) | |
| 16:10:18 | mordred | heh | |
| 16:10:25 | dtantsur | because that means we're still stuck with versioned endpoints everywhere | |
| 16:10:39 | dtantsur | aaaand microversion discovery also? | |
| 16:10:51 | mordred | yeah- we need to figure out when it is the correct thing to do to do the discovery logic even with an endpoint override | |
| 16:11:01 | dtantsur | man we need to dedicate a whole day in Berlin in November just to sit down and go through this mess | |
| 16:11:06 | mordred | for microversion discovery it will do the fetch on the endpoint override | |
| 16:11:12 | mordred | dtantsur: ++ | |
| 16:11:13 | mordred | seriouslyu | |
| 16:11:27 | mordred | because therea re like 12 competing use cases | |
| 16:11:32 | dtantsur | yep | |
| 16:12:22 | dtantsur | ideally, I'd prefer to endpoint-override to be unversioned (but support versioned ones too) | |
| 16:12:32 | dtantsur | my head hurts just from thinking about it.. | |
| 16:16:20 | mordred | dtantsur: yeah. such awesomeness | |
| 16:16:56 | dtantsur | the next two functions are also awesome: they handle the fact that some API use empty string instead of null :) | |
| 16:17:08 | dtantsur | which is not an issue for Python but a big issue for a statically typed language :D | |
| 16:32:18 | mordred | dtantsur|afk: ok. I think I have the versioned/unversioned thing handled | |