| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2020-06-07 | |||
| 14:40:40 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Import command group support from osc-lib https://review.opendev.org/733964 | |
| 14:40:41 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Add load_plugins option https://review.opendev.org/733961 | |
| 16:11:16 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Add load_plugins option https://review.opendev.org/733961 | |
| 16:11:16 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Import command group support from osc-lib https://review.opendev.org/733964 | |
| 16:17:11 | sshnaidm|afk | why can't I see from SDJK that server is locked? I get "locked: null" in any case | |
| 16:17:17 | sshnaidm|afk | *SDK | |
| 16:18:15 | sshnaidm|afk | is it a bug? ^^ if so, where should I report it? | |
| 16:20:09 | mordred | sshnaidm|afk: here is good - can you do openstack.enable_logging(http_debug) and do a fetch of the server so we can capture the http interaction to make sure we're getting the locked field back from the cloud? | |
| 16:21:24 | mordred | sshnaidm|afk: locked is New in version 2.9 - it's possible we're not negotiating for microversion properly | |
| 16:22:36 | mordred | there are older versions of nova that have a bug where the version discovery doc was auth-protected, which would cause this behavior. nova has been fixed - and keystoneauth has also been fixed to work around it - but if you have either an old nova or an old keystoneauth it can be an issue | |
| 16:23:44 | sshnaidm|afk | yeah, I don't see "lock" in requests at all.. | |
| 16:24:15 | sshnaidm|afk | it's old, but not sure how old | |
| 16:24:27 | mordred | can you try with most-recent keystoneauth? | |
| 16:24:45 | sshnaidm|afk | yeah, will try in CI | |
| 16:24:52 | mordred | (this might be a patch that would be valuable to backport to stable ksa if it fixes you) | |
| 16:25:46 | mordred | sshnaidm|afk: https://review.opendev.org/#/c/685042/ fwiw | |
| 16:27:13 | sshnaidm|afk | ack | |
| 17:13:54 | sshnaidm|afk | why does rebuild in SDK asks for an admin password? According to nova api it's not mandatory: https://docs.openstack.org/api-ref/compute/?expanded=rebuild-server-rebuild-action-detail#rebuild-server-rebuild-action | |
| 17:14:09 | sshnaidm|afk | if not specifying it will be generated | |
| 18:03:07 | openstackgerrit | Sagi Shnaidman proposed openstack/openstacksdk master: Make optional name and admin password https://review.opendev.org/733991 | |
| 18:03:12 | sshnaidm|afk | mordred, ^ | |
| 18:09:42 | openstackgerrit | Sagi Shnaidman proposed openstack/ansible-collections-openstack master: Add OpenstackModule to os_server_action https://review.opendev.org/721746 | |
| 18:12:37 | openstackgerrit | Sagi Shnaidman proposed openstack/ansible-collections-openstack master: Don't pass tenant_id for remote group https://review.opendev.org/730422 | |
| 18:18:06 | openstackgerrit | Sagi Shnaidman proposed openstack/ansible-collections-openstack master: Add OpenstackModule to os_server_action https://review.opendev.org/721746 | |
| 22:03:27 | openstackgerrit | Sagi Shnaidman proposed openstack/ansible-collections-openstack master: Add OpenstackModule to os_server_action https://review.opendev.org/721746 | |
| #openstack-sdks - 2020-06-08 | |||
| 01:55:57 | openstackgerrit | Monty Taylor proposed openstack/osc-lib master: Remove commandmanager subclass https://review.opendev.org/734000 | |
| 06:47:15 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Wait for the project cleanup to complete https://review.opendev.org/734016 | |
| 06:51:41 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Do not clean keypairs in the project cleanup https://review.opendev.org/734019 | |
| 09:00:11 | sshnaidm | cores, please take a look in your time https://review.opendev.org/#/c/733991/ | |
| 11:56:37 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Do not clean keypairs in the project cleanup https://review.opendev.org/734019 | |
| 12:50:24 | openstackgerrit | Sean McGinnis proposed openstack/keystoneauth master: Remove mox3 from local upper-constraints https://review.opendev.org/734071 | |
| 13:13:40 | openstackgerrit | Merged openstack/openstacksdk master: Make optional name and admin password https://review.opendev.org/733991 | |
| 14:15:06 | mordred | gtema__: if you get a sec, I started hacking on plugin loading in osc: https://review.opendev.org/#/q/topic:osc-plugin-optimize - right now it's just about unwinding the deps a little bit and optimizing a little bit - next I'll work on the client creation steps | |
| 14:15:17 | mordred | I think it shouldn't take too long to get those bits done | |
| 14:15:55 | mordred | however, looking at the osc-lib patch, I'm not convinced osc-functional-devstack and osc-functional-devstack-tips are doing the right things - I'd expect tips to pass and non-tips to fail there | |
| 14:16:15 | gtema | yeah, I'm fighting with the project cleanup, which seems also not that easy wrt filters (dry_run with filters is pretty much useless in the network area) | |
| 14:17:26 | gtema | you can't easily estimate, whether ports would be present in the real run or not (re-evaluate whether dropping resource will result in dropping port or not) | |
| 14:19:59 | gtema | mordred: what is your intention is "black"? | |
| 14:20:26 | mordred | gtema: dunno. that was just me playing | |
| 14:20:32 | gtema | okay | |
| 14:20:58 | mordred | gtema: good point re: ports - thats "fun" :( | |
| 14:21:03 | gtema | some things are indeed interesting, but replacing ' to " seems really mad | |
| 14:21:16 | openstackgerrit | Monty Taylor proposed openstack/python-openstackclient master: Add cliff to libs-from-git for devstack functional tips https://review.opendev.org/734095 | |
| 14:21:43 | gtema | I think we also need to have one src job in the osc with all it deps | |
| 14:21:52 | mordred | gtema: yeah. that one is a bit annoying. I like the indentation choice - I think it's good for making refactoring patches cleaner | |
| 14:22:07 | gtema | agree | |
| 14:22:47 | nightmare_unreal | frickler: hey , if you get time. Can you read my comment here : https://review.opendev.org/#/c/733573/ . Basically it's regarding that it's cliff that handles the output formatting . So the problem is not in osc-client per se ( I think) | |
| 14:22:50 | mordred | gtema: and yes - the osc patch there updates the tips job to include cliff | |
| 14:23:35 | gtema | we need same for sdk, osc-lib and cliff | |
| 14:24:02 | gtema | we will have pretty much changes soon and it becomes madness | |
| 14:26:34 | openstackgerrit | Monty Taylor proposed openstack/python-openstackclient master: Add cliff to libs-from-git for devstack functional tips https://review.opendev.org/734095 | |
| 14:26:35 | openstackgerrit | Monty Taylor proposed openstack/python-openstackclient master: Remove os-client-config references https://review.opendev.org/734097 | |
| 14:27:22 | mordred | gtema: when you say "need same" - what do you mean? you mean updating the jobs there to make sure it's all good? | |
| 14:27:37 | mordred | also - there's a "fun" thing about osc and devstack jobs that I don't think we've fully captured in our test jobs yet | |
| 14:28:17 | gtema | oh, I missed definition of the functional-devstack-tips. I thought it was not taking libs from git, but only release ones | |
| 14:28:38 | mordred | we have "release vs src" - and we have that for both devstack/openstack itself, as well as the client stuff installed into the tox virtualenv | |
| 14:28:56 | gtema | yes, now I see | |
| 14:44:00 | frickler | nightmare_unreal: well if you think you can fix this in cliff, go ahead. | |
| 14:50:52 | dtantsur | mordred: hey, do you remember from the top of your head, if OS_AUTH_URL has /v3/ in the end, do we trim it in keystoneauth? | |
| 14:50:57 | dtantsur | this is re https://github.com/dtantsur/rust-openstack/issues/114 | |
| 14:51:29 | gtema | in my case it has v3 | |
| 14:51:53 | mordred | dtantsur: we don't trim it if we're looking for version 3 | |
| 14:52:21 | mordred | dtantsur: I think the only time we trim it and use unversioned is if someone requests v2 | |
| 14:52:29 | dtantsur | gotcha | |
| 14:52:39 | mordred | because /v3 should also contain a versioned discovery doc | |
| 14:53:48 | dtantsur | https://github.com/dtantsur/rust-osauth/blob/master/src/identity.rs#L226 doesn't account for a trailing slash #facepalm | |
| 14:54:06 | mordred | dtantsur: hah | |
| 14:54:15 | dtantsur | this project has a loong way ahead in terms of robustness | |
| 14:54:22 | dtantsur | on the other hand, people are using it \o/ | |
| 14:54:25 | mordred | dtantsur: this comes back to the ptg discussion about trailing vs non-trailing in catalogs | |
| 14:54:27 | mordred | dtantsur: \o/ | |
| 14:54:38 | mordred | I pinged sdague about that btw and he did not remember | |
| 14:55:08 | dtantsur | too much time has passed | |
| 14:55:48 | mordred | yah. I remember having a summit discussion at some summit. I mostly remember that the room was downstairs | |
| 14:55:59 | mordred | but I don't remember which city ... maybe it was atlanta? | |
| 14:56:17 | mordred | might have been paris | |
| 14:56:31 | mordred | or hong kong | |
| 14:56:47 | gtema_ | :D | |
| 15:33:06 | openstackgerrit | Merged openstack/ansible-collections-openstack master: Add setup.py for install with pip https://review.opendev.org/730400 | |
| 15:34:47 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Add load_plugins option https://review.opendev.org/733961 | |
| 15:36:15 | openstackgerrit | Monty Taylor proposed openstack/osc-lib master: Remove commandmanager subclass https://review.opendev.org/734000 | |
| 15:38:46 | mordred | gtema: sorry - I missed this one in the topic: https://review.opendev.org/#/c/733968/1 | |
| 15:39:42 | gtema | done | |
| 15:39:57 | gtema | feel free to set +A yourself | |
| 15:40:26 | mordred | gtema: thansk! | |
| 15:41:38 | mordred | gtema: so - with the cliff patch, the idea is that in osc we'll have a file that has a ton of "commandmanager.add_command" calls pointing to the classes in-tree rather than registering our own plugins as entrypoints | |
| 15:42:00 | mordred | then, since cliff will first attempt to resolve those from what it knows before scanning entrypoints | |
| 15:42:23 | mordred | for the common case we can avoid the entrypoint cost- and only take that cost if someone asks for a command tha we don't know | |
| 15:42:36 | gtema | yupp, makes sense | |
| 16:20:38 | openstackgerrit | Monty Taylor proposed openstack/cliff master: Add load_plugins option https://review.opendev.org/733961 | |
| 16:40:14 | smcginnis | mordred: Hey there. Doing a search on requirements, I can't find any repos that have shade declaring it as a dependency in their requirements files. Does that seem right? | |
| 16:40:43 | mordred | smcginnis: yeah - I don't think anything should be depending on shade at this point | |
| 16:40:58 | mordred | smcginnis: everything that used to should be depending on openstacksdk at this point | |
| 16:41:01 | mordred | smcginnis: same with os-client-config | |
| 16:41:09 | smcginnis | OK, great. Another one I can clean out of global reqs then. | |
| 16:41:28 | mordred | smcginnis: we're probably in a good place to change the release model for them now too | |
| 16:41:39 | mordred | smcginnis: maybe just to independent? | |
| 16:42:11 | smcginnis | mordred: If you think it will only need occasional releases from time to time, that may be better. | |
| 16:44:09 | mordred | smcginnis: yeah- the're mostly on life-support for backwards compat | |
| 16:44:17 | mordred | they definitely don't need cycle based releases | |