Wednesday, 2025-10-01

*** mhen_ is now known as mhen01:11
*** tkajinam_ is now known as tkajinam10:54
*** darmach47 is now known as darmach13:01
d34dh0r53#startmeeting keystone15:07
opendevmeetMeeting started Wed Oct  1 15:07:00 2025 UTC and is due to finish in 60 minutes.  The chair is d34dh0r53. Information about MeetBot at http://wiki.debian.org/MeetBot.15:07
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:07
opendevmeetThe meeting name has been set to 'keystone'15:07
d34dh0r53Reminder: This meeting takes place under the OpenInfra Foundation Code of Conduct15:07
gtemaoh, here are the folks15:07
d34dh0r53#link https://openinfra.dev/legal/code-of-conduct15:07
xeko/15:07
d34dh0r53#topic roll call15:07
gtemao/15:07
d34dh0r53admiyo, bbobrov, crisloma, d34dh0r53, dpar, dstanek, hrybacki, lbragstad, lwanderley, kmalloc, rodrigods, samueldmq, ruan_he, wxy, sonuk, vishakha, Ajay, rafaelwe, xek, gmann, zaitcev, reqa, dmendiza[m], dmendiza, mharley, jph, gtema, cardoe, deydra15:08
d34dh0r53o/15:08
d34dh0r53sorry for missing the reviewathon last week, I had to take a personal day15:08
gtemanp15:08
dmendiza[m]🙋‍♂️15:11
d34dh0r53#topic review past meeting work items15:11
d34dh0r53#link meetings.opendev.org/meetings/keystone/2025/keystone.2025-08-27-15.04.html15:12
d34dh0r53looks like we haven't had a meeting since August :o15:12
d34dh0r53gtema and I have one action item, which I'll take care of today15:12
d34dh0r53 dwilde/gtema add PTG topic about service account15:12
d34dh0r53although I'm not sure of the context15:13
d34dh0r53that was about 100 years ago in my brain ;)15:13
gtemaI know - will do now15:13
gtemaadded agenda to etherpad15:14
gtemasince we are on that we can discuss the times for keystone meetings15:14
d34dh0r53yeah, I booked on Tues and Wed just to hold a room but we can definitely move things around15:15
d34dh0r53#topic ptg planning15:15
gtemaah ok, i missed that15:16
gtemacould stay like that - for me it's perfectly fine15:16
dmendiza[m]#link https://ptg.opendev.org/ptg.html15:17
d34dh0r53cool, thanks dmendiza 15:17
d34dh0r53we also need to have a combined session with the horizon team15:18
d34dh0r53I can reach out to them to figure out the best time for that15:18
d34dh0r53#action dwilde plan PTG session with Horizon folks15:18
gtemacool, thks15:19
d34dh0r53#topic liaison updates15:20
d34dh0r53nothing from me15:20
gtemaneither me15:21
d34dh0r53#topic specification OAuth 2.0 (hiromu)15:21
d34dh0r53#link https://review.opendev.org/q/topic:bp%252Foauth2-client-credentials-ext15:21
d34dh0r53#link https://review.opendev.org/q/topic:bp%252Fenhance-oauth2-interoperability15:21
d34dh0r53no updates from me on this15:21
d34dh0r53#topic specification Secure RBAC (dmendiza)15:21
d34dh0r53#link https://governance.openstack.org/tc/goals/selected/consistent-and-secure-rbac.html#z-release-timeline_15:21
d34dh0r532025.2 Release Timeline15:21
d34dh0r53Update oslo.policy in keystone to enforce_new_defaults=True15:21
d34dh0r53Update oslo.policy in keystone to enforce_scope=True15:21
dmendiza[m]🙋‍♂️15:22
dmendiza[m]First day back from PTO, so I have no idea what's going on. 😅15:22
dmendiza[m]I think we still have some code in devstack that is defaulting to false15:23
* dmendiza[m] checks gerrit15:23
gtema2025.2 is gone as a target today15:23
gtemaso its from now 2026.115:23
dmendiza[m]ack, we need to update the agenda 15:24
dmendiza[m]#link https://review.opendev.org/c/openstack/devstack/+/95621015:24
dmendiza[m]^^ I just rebased that.  I think oslo.policy is going to remove the options soon.15:25
gtemayeah, should do that sooner rather than later15:25
dmendiza[m]It failed the gate last time, so I'll keep an eye on the rebase and try to fix whatever comes up15:26
d34dh0r53thanks dmendiza 15:26
d34dh0r53#topic specification OpenAPI support (gtema)15:26
d34dh0r53#link https://review.opendev.org/q/topic:%22openapi%22+project:openstack/keystone15:26
gtemaall crear, no changes. But I am from time to time struggling in development15:27
gtemaI mean doing certain experiments I end up with "updated" data in the db and json schema validation fail15:27
gtemathis result in no entries returned at all - this sucks15:27
gtemanature of such experiments is the keystone-ng15:28
gtemarust code returns such entries, but python not. I need to think again on the conceptual level about the topic15:29
d34dh0r53yep, that does suck15:29
d34dh0r53ack, thanks gtema 15:29
dmendiza[m]Dave Wilde (d34dh0r53): I added a topic to the agenda15:29
d34dh0r53thanks dmendiza 15:30
d34dh0r53#topic open discussion15:30
d34dh0r53drencrom15:30
d34dh0r53pep8 (mypy) is broken on 2024.2 branch (see for example https://zuul.opendev.org/t/openstack/build/2fdbd3164c8c4241a5a6edd1895f6d3c)15:30
gtemathis should have been fixed some weeks ago15:30
dmendiza[m]🙋‍♂️ > Secuirty Compliance Testing (dmendiza)15:30
drencromhi, I think there might be some depencies issues on that branch but could not fix it myself15:31
gtemaoh, actually no. I guess I have not backported the fix so far away15:31
gtemahttps://review.opendev.org/c/openstack/keystone/+/958665 should be backported most likely15:32
gtemayes, it is the same issue15:33
opendevreviewArtem Goncharov proposed openstack/keystone stable/2024.2: Ignore typing on the single import  https://review.opendev.org/c/openstack/keystone/+/96269615:34
gtemaI wanted us to have a short chat about https://bugs.launchpad.net/keystone/+bug/212261515:35
drencromOk, I'll add a depends tag on that one and see it if works15:35
gtemaI was riping my hair out of my head tracing where the revocation and validation happens - it is a spaghetti with multiple tabs15:36
gtemaanyway: any objections for us doing what I proposed in the last comment or should we rather in such case explicitly generate revokation15:37
gtemathe point is that I have found unit test that checks that token for the disabled user does not work even after re-enabling it back (through the revocation entry)15:37
gtemaif we only check additionally whether the user is now active or not to address the mentioned usecase the token will be again accepted once the user is activated again what is a different flow compared to what is the "normal" behavior15:38
gtemaon the other hand ... it all sucks since we do not really have any info of user that got disabled in a remote backend15:39
dmendiza[m]was catching up on the bug thread15:45
dmendiza[m]> when validating a token ensure the user is active or disabled15:47
dmendiza[m]^^ that suggestion, right?15:47
gtemayes - reject the token when the user status is disabled15:48
dmendiza[m]so the question is whether to invalidate the token or not?  15:48
gtemaplace in code is pretty straight forward. Just the question: simply reject the token or generate a revoke event similar to what is done when user is disabled through api15:49
gtemait's not actually a token revocation - it is a user revocation15:49
dmendiza[m]Disabling the user if we see that the backend status is disabled seems like the right thing to do.15:50
gtemahttps://opendev.org/openstack/keystone/src/commit/4275c6801e165f400d9bfadf4ea37e4a6b115e3f/keystone/identity/core.py#L137215:50
gtemawhen now the user enabled attr is changed the revocation event for the user is generated that ensures that even after reenabling the old token work15:51
opendevreviewJorge Merlino proposed openstack/keystone stable/2024.2: Fix AD nested groups issues  https://review.opendev.org/c/openstack/keystone/+/96141315:51
gtemadmendiza - we do not disable the user explicitly - that's the point of the bug15:52
dmendiza[m]Yeah, I think we agree that we should:15:52
dmendiza[m]* invalidate token15:53
dmendiza[m]* disable user15:53
gtemawhen a backend driver is used and the user entry is changed there bypassing the api (keycloak, ldap) we do not have any info on that, but once we fecth user detail or it tries to login we see updated into15:53
gtemawe can't invalidate the token because we do not know all the tokens the user has - thus need to add additional validation that bypasses regular revocation15:54
gtemaok, we are stretching time with that too much15:55
d34dh0r53Yeah, but that segues into the bug review15:55
d34dh0r53#topic bug review15:56
d34dh0r53#link https://bugs.launchpad.net/keystone/?orderby=-id&start=015:56
d34dh0r53two bugs for keystone, the first we just discussed15:56
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/212261515:56
d34dh0r53next is15:56
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/212504215:56
gtemathat looks like a duplicate of what recently got merged15:57
gtemahttps://bugs.launchpad.net/keystone/+bug/210209615:58
gtemalaunchpad does not have duplicate option???16:00
d34dh0r53I thought it did16:01
gtemaI can't find that16:02
gtemaanyway - you can go on with next ones16:02
d34dh0r53huh, I guess wont-fix with a note and link to the dup16:02
d34dh0r53anyways16:02
d34dh0r53#link https://bugs.launchpad.net/python-keystoneclient/?orderby=-id&start=016:02
d34dh0r53nothing new here16:03
d34dh0r53#link https://bugs.launchpad.net/keystoneauth/+bugs?orderby=-id&start=016:03
d34dh0r53keystoneauth is good16:03
d34dh0r53#link https://bugs.launchpad.net/keystonemiddleware/+bugs?orderby=-id&start=016:03
d34dh0r53no new bugs in keystonemiddleware16:03
d34dh0r53#link https://bugs.launchpad.net/pycadf/+bugs?orderby=-id&start=016:03
d34dh0r53pycadf is good16:04
d34dh0r53#link https://bugs.launchpad.net/ldappool/+bugs?orderby=-id&start=016:04
d34dh0r53so is ldappool16:04
d34dh0r53#topic conclusion16:04
d34dh0r53add topics to the ptg etherpad16:04
gtemaservice accounts topic is there already16:04
d34dh0r53#link https://etherpad.opendev.org/p/oct2025-ptg-keystone16:05
d34dh0r53thanks gtema 16:05
d34dh0r53that's all from me16:05
gtemathanks Dave Wilde (d34dh0r53) , have a nice day16:06
d34dh0r53thanks, likewise :)16:06
d34dh0r53#endmeeting16:06
opendevmeetMeeting ended Wed Oct  1 16:06:30 2025 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)16:06
opendevmeetMinutes:        https://meetings.opendev.org/meetings/keystone/2025/keystone.2025-10-01-15.07.html16:06
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/keystone/2025/keystone.2025-10-01-15.07.txt16:06
opendevmeetLog:            https://meetings.opendev.org/meetings/keystone/2025/keystone.2025-10-01-15.07.log.html16:06
opendevreviewJorge Merlino proposed openstack/keystone stable/2024.2: Fix AD nested groups issues  https://review.opendev.org/c/openstack/keystone/+/96141318:51
cardoegtema: I'll be interested in your service account convo. Currently trying to reconcile that with federated logins as well. e.g. OIDC really only wants to have client credentials without userinfo21:45

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!