| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-03-08 | |||
| 05:26:52 | adriant | 0.11.3 was the last one | |
| 05:27:07 | adriant | 0.12.0 is 2 days old :P | |
| 05:27:16 | eandersson | Keep hitting obscure incompatibility errors that get eaten up | |
| 05:27:30 | eandersson | > name = urllib.parse.urlparse(authenticator.auth_url).hostname | |
| 05:27:30 | eandersson | The worst one is from this line | |
| 05:28:12 | eandersson | Many code paths lead here with authenticator being set to None | |
| 05:29:56 | eandersson | Since before 0.10, authenticator was set automatically | |
| 05:31:50 | eandersson | Actually, think I will just pass on the authenticator for now | |
| 05:32:17 | eandersson | Since I feel like the senlinclient needs to be fixed (or cleaned up) before the new route makes snese | |
| 05:32:53 | eandersson | I assume I can just pass on a keystoneauth1 Token right? | |
| 05:43:01 | adriant | mordred: nvm... I'm silly. I can just do: role_id="role.id" for some reason I suddenly thought it was the other way around... | |
| 05:48:54 | eandersson | This works with both openstacksdk 0.9.x and 0.10.x http://paste.openstack.org/show/694585/ | |
| 05:49:11 | eandersson | adriant, does this make sense? ^ | |
| 05:53:25 | openstackgerrit | Adrian Turjak proposed openstack/python-openstacksdk master: Redo role assignment list query filters https://review.openstack.org/550700 | |
| 05:53:38 | adriant | eandersson: and fails with 0.11.3+ ? | |
| 05:54:52 | eandersson | Sorry I meant that it works with 0.9.x and 0.10.x and 0.11.x | |
| 05:55:04 | eandersson | The previous code did not, as there was no authenticator passed on | |
| 05:55:41 | adriant | eandersson: what module is senlin_client exactly? | |
| 05:55:48 | adriant | trying to find it in the senlinclient code | |
| 05:56:15 | adriant | eandersson" https://github.com/openstack/python-senlinclient/blob/master/senlinclient/v1/client.py#L18 ? | |
| 05:56:22 | eandersson | Yep | |
| 05:56:31 | eandersson | https://github.com/openstack/python-senlinclient/blob/master/senlinclient/plugin.py#L29 | |
| 05:56:46 | eandersson | Basically it still uses the deprecated profile | |
| 05:56:59 | eandersson | the problem with profile after 0.10 and newer is that you now need to pass an authenticator | |
| 05:57:19 | adriant | oh, but if you pass through the authenticator as a kwarg it works | |
| 05:57:25 | adriant | yeah, that makes sense | |
| 05:57:45 | eandersson | I feel like the authenticator should still be passed on until profile has been removed :p | |
| 05:57:56 | eandersson | I mean for the openstacksdk | |
| 05:59:12 | adriant | eandersson: so wait... the senlin client uses the cluster service in the sdk? | |
| 05:59:21 | adriant | why not just use the SDK directly? | |
| 05:59:33 | eandersson | good question | |
| 05:59:40 | eandersson | probably legacy code :D | |
| 05:59:59 | adriant | if the SDK is missing logic that the senlin client has in it... then we should move that over | |
| 06:00:05 | eandersson | Yea for sure | |
| 06:00:23 | adriant | and in your case doing the dashboard stuff you're better off using the sdk directly | |
| 06:00:27 | eandersson | My immediate concern is to fix the queens release thou :p | |
| 06:00:28 | adriant | unless, again the client is missing stuff | |
| 06:00:32 | adriant | Oh | |
| 06:00:33 | adriant | kk | |
| 06:01:57 | eandersson | Also, we should get dns on the sdk :D | |
| 06:02:10 | adriant | OH, hah, yep the senlinclient is literally a proxy to the SDK | |
| 06:02:24 | adriant | so the only code the senlinclient adds is the CLI stuff | |
| 06:02:32 | eandersson | https://review.openstack.org/#/c/550701/ | |
| 06:07:12 | eandersson | but yea should just call out the openstacksdk directly | |
| 06:07:47 | adriant | I'm trying to switch all my code to the SDK client, hence why I keep finding all kind of weird little bugs :P | |
| 06:09:48 | adriant | that said, eandersson, you should be able to still pass in auth variables rather than an authenticator, it's just that the variables for auth need to be in an auth param. | |
| 06:11:09 | eandersson | Do you mean instead of generic.Token? | |
| 06:12:09 | adriant | lemme try something and send you code snippet to try | |
| 06:12:25 | eandersson | The problem is this code https://github.com/openstack/python-openstacksdk/blob/0.11.3/openstack/profile.py#L50 | |
| 06:12:53 | eandersson | https://github.com/openstack/python-openstacksdk/blob/0.11.3/openstack/connection.py#L277 | |
| 06:13:01 | eandersson | I would appreciate that adriant | |
| 06:18:21 | adriant | eandersson: hmmm, i think you're right, until that profile logic goes away, you have to pass in the authenticator | |
| 06:19:02 | eandersson | Wish that portion would have been kept for backwards compatibility | |
| 06:35:39 | openstackgerrit | Olivier Bourdon proposed openstack-infra/shade master: Add extra failure codes to bad request exception https://review.openstack.org/550107 | |
| 08:24:06 | openstackgerrit | Tobias Brox proposed openstack-infra/shade master: Add subnetpool_name_or_id parameter to create_subnet https://review.openstack.org/548722 | |
| 09:55:54 | openstackgerrit | Olivier Bourdon proposed openstack-infra/shade master: Add proper return value for validate_node https://review.openstack.org/549981 | |
| 11:17:33 | hwoarang | hello. we are seeing the following problems with queens https://bugs.launchpad.net/python-openstacksdk/+bug/1754302 not sure if anyone has any suggestion besides downgrading the openstacksdk component :) | |
| 11:17:34 | openstack | Launchpad bug 1754302 in OpenStack SDK "openstacksdk-0.12.0 breaks with self-signed certificates" [Undecided,New] | |
| 11:30:01 | openstackgerrit | Lingxian Kong proposed openstack/service-types-authority master: Add qinling project https://review.openstack.org/550787 | |
| 12:15:11 | openstackgerrit | James Denton proposed openstack/python-openstacksdk master: Adds prefixlen to the request body when creating subnets https://review.openstack.org/550558 | |
| 12:41:30 | cdent | dtantsur, elmiko, edleafe: can one of you run the meeting today please? I've cut my thumb somewhat badly (but not catastrophically) and typing is slow | |
| 12:42:08 | dtantsur | cdent: ugh, get better :( yes, I can | |
| 12:42:43 | cdent | thanks | |
| 12:53:55 | mordred | adriant: oh wow. I see much scrollback | |
| 13:02:05 | elmiko | cdent: sorry to hear about the thumb =( | |
| 13:02:27 | elmiko | cdent dtantsur edleafe , i will most likely miss the meeting today, i've got a conflict internally here | |
| 13:31:39 | edleafe | cdent: I'll run it | |
| 13:46:00 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Remove tox_install.sh and align with constraints consumption https://review.openstack.org/550837 | |
| 14:01:54 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Remove tox_install.sh and align with constraints consumption https://review.openstack.org/550837 | |
| 16:48:03 | edleafe | dtantsur: cdent: redy for review https://etherpad.openstack.org/p/api-sig-newsletter | |
| 16:48:10 | edleafe | geez | |
| 16:48:12 | edleafe | *ready | |
| 16:48:47 | dtantsur | short and nice, lgtm :) | |
| 16:51:08 | cdent | edleafe: you wanna add the cache-control review? | |
| 17:01:04 | edleafe | cdent: ah, good catch | |
| 17:03:18 | edleafe | added and shipped | |
| 17:08:46 | cdent | woot | |
| 17:57:38 | lbragstad | o/ curious if anyone from the api sig/wg would be interesting in sharing their thoughts on http://lists.openstack.org/pipermail/openstack-dev/2018-March/128027.html ? | |
| 17:57:44 | lbragstad | interested* | |
| 17:58:20 | lbragstad | the team is curious about where we should meet usability with the consistency of the rest of the API | |
| 17:58:43 | cdent | lbragstad: will put that in my queue for tomorrow. edleafe, dtantsur|afk ^ | |
| 17:59:15 | lbragstad | thanks cdent | |
| 17:59:46 | cdent | would today but I got melty brain | |
| 17:59:56 | lbragstad | that's understandable | |
| 18:00:25 | lbragstad | s/coffee/beer/ if symptoms persist :) | |
| 18:02:40 | cdent | :) | |
| 18:28:47 | edleafe | wait - brains aren't supposed to be melty? | |
| 19:42:48 | mordred | edleafe: oy. I've been doing it all wrong | |
| 21:41:13 | adriant | mordred: I found a bug, although this one is easy: https://review.openstack.org/#/c/550700/ | |
| 21:41:38 | adriant | I also took the chance to check all our resources for identity and make sure the query filters matched. They did, just this one was wrong. | |
| 21:42:34 | adriant | lbragstad: ^ I you fellows are evil for "user.id" as the filter rather than "user_id" :P | |
| 21:42:57 | adriant | and I think it's the only keystone API to do that. :( | |
| 21:43:43 | lbragstad | hmmmmmm | |
| 21:43:56 | lbragstad | i might have to dig into why that is | |
| 21:45:47 | adriant | lbragstad: https://developer.openstack.org/api-ref/identity/v3/#id595 this one | |
| 21:46:25 | adriant | I always used the keystoneclient so never realised until now when trying to switch to the sdk that the API was weird :P | |
| 21:47:25 | adriant | the keystoneclient handles it here: https://github.com/openstack/python-keystoneclient/blob/master/keystoneclient/v3/role_assignments.py#L101-L122 | |
| 22:18:47 | mordred | adriant: that's ... that's just amazing | |
| 22:19:18 | lbragstad | "you had *one* job!" | |
| 22:19:25 | mordred | :) | |