15:02:24 <d34dh0r53> #startmeeting keystone 15:02:24 <opendevmeet> Meeting started Wed May 29 15:02:24 2024 UTC and is due to finish in 60 minutes. The chair is d34dh0r53. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:24 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:24 <opendevmeet> The meeting name has been set to 'keystone' 15:02:35 <d34dh0r53> #topic roll call 15:02:46 <d34dh0r53> admiyo, bbobrov, crisloma, d34dh0r53, dpar, dstanek, hrybacki, knikolla[m], lbragstad, lwanderley, kmalloc, rodrigods, samueldmq, ruan_he, wxy, sonuk, vishakha, Ajay, rafaelwe, xek, gmann, zaitcev, reqa, dmendiza[m], mharley, jph, gtema 15:02:46 <gtema> o/ 15:02:47 <mharley[m]> o/ 15:02:48 <d34dh0r53> o/ 15:04:10 <d34dh0r53> #topic review past meeting work items 15:04:17 <d34dh0r53> #link https://meetings.opendev.org/meetings/keystone/2024/keystone.2024-05-22-15.01.html 15:04:51 <d34dh0r53> I've been thinking about my action item and I think it's best to handle those things with release notes and not have a known issues as I don't think anyone has the bandwidth to maintain it 15:05:26 <d34dh0r53> So...without any objections I'll punt on the known issues section 15:06:12 <gtema> wfm 15:06:30 <d34dh0r53> cool 15:06:49 <d34dh0r53> moving on 15:07:21 <d34dh0r53> #topic liaison updates 15:07:52 <d34dh0r53> nothing from VMT or releases 15:08:49 <d34dh0r53> #topic specification OAuth 2.0 (hiromu) 15:08:59 <d34dh0r53> #link https://review.opendev.org/q/topic:bp%252Foauth2-client-credentials-ext... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/kdtXgbfJHgyeUToLyMzvoNEp>) 15:09:51 <d34dh0r53> I'm going to work on rebasing the last outstanding patch and see how it goes 15:09:58 <d34dh0r53> next up we have 15:10:15 <d34dh0r53> #topic specification Secure RBAC (dmendiza[m])... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/pHxYbHZRGakynbkUtTdYMcJo>) 15:11:38 <d34dh0r53> dmendiza: you around? 15:13:00 <dmendiza[m]> 🙋 15:13:16 <dmendiza[m]> I don't htink we have any SRBAC udpates this week? 15:13:26 <dmendiza[m]> Maybe just the tempest-plugin 0.16.0 release? 15:14:06 <d34dh0r53> Yeah, we released a new version of keystone-tempest-plugin last week, it contains the latest SRBAC testing fixes 15:15:10 <dmendiza[m]> #link https://releases.openstack.org/dalmatian/index.html#keystone-tempest-plugin 15:17:57 <d34dh0r53> awesome, thanks dmendiza ! 15:18:00 <d34dh0r53> next up 15:18:16 <d34dh0r53> #topic specification Improve federated users management (gtema) (on-hold until 05/29) 15:18:16 <d34dh0r53> #link https://review.opendev.org/c/openstack/keystone-specs/+/748748 15:18:16 <d34dh0r53> gtema: revoke support for the approach 15:18:32 <gtema> this one is going to be long 15:18:46 <gtema> I was thinking long time, prototyping, modelling 15:19:01 <gtema> talking also to mnasser on his tour across Europe 15:19:35 <gtema> and came to conclusion, that the approach is wrong. You should not try to model roles/projects/users/groups/domains mapping outside of Keystone 15:19:48 <gtema> and leave IdP to take care of users and groups 15:20:09 <gtema> with https://github.com/vexxhost/keystone-keycloak-backend it is very easy to get users visible in Keystone 15:20:53 <gtema> and once user is disabled in Keycloak (external IdP) it becomes immediately applied on Keystone side and user is not able to continue accessing cloud i.e. using ApplicationCredentials 15:21:11 <gtema> note: this is not the case with ephemeral users 15:21:17 <gtema> BUT 15:21:42 <gtema> Keystone docs recommend to create out-of tree drivers for cases like that - all good 15:21:54 <gtema> the problem comes with configuration of out-of-tree backend drivers 15:22:12 <gtema> option a) FS (as domain_specific configs) 15:22:41 <gtema> this option works great except that it requires restart of keystone once you add new domain - this is a no go 15:22:51 <gtema> option b) DB 15:23:18 <gtema> here there is a problem that current implementation does not allow having drivers their own configuration 15:23:43 <gtema> and more then that, the API of domain_configs is hardcoding options that can be set 15:23:58 <gtema> so now let's come to questions: 15:24:24 <gtema> what is the preferred/recommended way dealing with config here: FS or DB? 15:25:00 <gtema> for the reference: https://docs.openstack.org/keystone/latest/admin/configuration.html#domain-specific-configuration 15:26:49 <gtema> Statement from docs: "Unlike the file-based method of specifying domain-specific configurations, options specified via the Identity API will become active without needing to restart the keystone server." 15:27:20 <d34dh0r53> Right, may have to wait for the cache to expire but no restart needed 15:27:51 <gtema> right 15:28:15 <gtema> from my pov using FS for config is not really an option for out-of-tree drivers. That leaves only DB 15:28:34 <gtema> and here I need to change 2 things in Keystone: 15:29:09 <gtema> a) allow out-of-tree drivers to register custom options (so that at least Keystone doesn't crash once I place them into DB) 15:29:38 <gtema> b) extend config to allow adding new driver specific options into the comain_config api 15:30:06 <gtema> https://opendev.org/openstack/keystone/src/branch/master/keystone/resource/core.py#L1048 15:30:21 <gtema> the link points to the place where current options are hardcoded 15:30:41 <gtema> fixing a) is something like 5 lines of code 15:30:56 <gtema> fixing b) seems also relatively strait forward 15:31:07 <gtema> BUT I need to know whether you are fine with that 15:31:48 <gtema> basically docs suggest people should go for out-of-tree drivers, but current code base makes it terrible hard to configure such drivers in such case 15:33:17 <d34dh0r53> I'm fine with it, I think keystone is moving in this direction and making the configuration easier for out of tree drivers is a win-win 15:33:39 <gtema> perfect. Then I' polish my changes in next days 15:33:58 <d34dh0r53> Excellent! 15:34:14 <d34dh0r53> Thank you gtema (Artem Goncharov) ! 15:34:18 <gtema> wlcm. 15:34:54 <gtema> I am going to write better docs on how to establish federation for a real life once everything stabilizes 15:35:08 <d34dh0r53> sweet 15:35:16 <gtema> and wrt the point on the tracker: I am pulling off my support for the mentioned spec 15:35:27 <gtema> and instead focus on this alternative way 15:35:59 <d34dh0r53> okay 15:36:06 <gtema> that's it from me on the topic 15:36:19 <d34dh0r53> thanks! 15:36:22 <d34dh0r53> next up 15:36:36 <d34dh0r53> #topic specification OpenAPI support (gtema) 15:36:36 <d34dh0r53> #link https://review.opendev.org/c/openstack/keystone-specs/+/910584 15:36:36 <d34dh0r53> gtema: waiting for reviews 15:37:00 <gtema> waitign for the spec to land. The one for Nova was merged recently 15:37:37 <d34dh0r53> cool, dmendiza , Grzegorz Grasza ^ can y'all review? 15:38:27 <d34dh0r53> #topic open discussion 15:38:37 <d34dh0r53> 'passlib update 15:38:51 <d34dh0r53> no update from me, I'm going to pin in the upper-requirements for now 15:39:12 <d34dh0r53> domain manager (mhen)... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/LkRBdKyQlgcNtIMSmDMIFiWc>) 15:39:18 <d34dh0r53> This needs reviews 15:40:04 <gtema> i restored my +1, last changes are polishing phrasing 15:40:51 <gtema> would be really great to land this one soon, since it is also a one dependency of having federation to external IdP 15:42:18 <d34dh0r53> indeed 15:43:00 <d34dh0r53> domain list scoping fix (mhen)... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/dgBmgIPIkmxSDHRtBPdfUGQf>) 15:43:23 <d34dh0r53> I think this was on hold as well, can I remove the WIP? 15:44:44 <d34dh0r53> I don't think mhen is around today, so we'll move on for the sake of time 15:44:47 <d34dh0r53> Enforcing scope in keystone breaks heat (and probably magnum) (tkajinam)... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/xPvMPjeJnSzhIGXjRjozjnPJ>) 15:45:39 <d34dh0r53> I think this is mostly done, just a couple more reviews and perhaps backporting of 916707 15:47:01 <d34dh0r53> cool, moving on 15:47:17 <d34dh0r53> unless anyone has something not on the agenda for open discussion 15:47:24 <gtema> not from me 15:47:31 <d34dh0r53> #topic bug review 15:47:41 <d34dh0r53> #link https://bugs.launchpad.net/keystone/?orderby=-id&start=0 15:48:11 <d34dh0r53> no new bugs for keystone 15:48:14 <d34dh0r53> #link https://bugs.launchpad.net/python-keystoneclient/?orderby=-id&start=0 15:48:46 <d34dh0r53> nothing new for python-keystoneclient 15:48:49 <d34dh0r53> #link https://bugs.launchpad.net/keystoneauth/+bugs?orderby=-id&start=0 15:49:13 <d34dh0r53> no new bugs for keystoneauth 15:49:16 <d34dh0r53> #link https://bugs.launchpad.net/keystonemiddleware/+bugs?orderby=-id&start=0 15:49:35 <d34dh0r53> keystonemiddleware is good 15:49:36 <d34dh0r53> #link https://bugs.launchpad.net/pycadf/+bugs?orderby=-id&start=0 15:49:54 <d34dh0r53> pycadf doesn't have any new issues 15:50:08 <d34dh0r53> #link https://bugs.launchpad.net/ldappool/+bugs?orderby=-id&start=0 15:50:15 <d34dh0r53> nor does ldappool 15:50:21 <d34dh0r53> #topic conclusion 15:50:29 <d34dh0r53> Nothing from me, thanks all! 15:50:49 <gtema> thanks 15:50:55 <d34dh0r53> #endmeeting