Wednesday, 2026-02-04

*** mhen_ is now known as mhen02:30
*** debian is now known as Guest146312:21
d34dh0r53#startmeeting keystone15:01
opendevmeetMeeting started Wed Feb  4 15:01:45 2026 UTC and is due to finish in 60 minutes.  The chair is d34dh0r53. Information about MeetBot at http://wiki.debian.org/MeetBot.15:01
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:01
opendevmeetThe meeting name has been set to 'keystone'15:01
d34dh0r53Reminder: This meeting takes place under the OpenInfra Foundation Code of Conduct15:01
d34dh0r53#link https://openinfra.dev/legal/code-of-conduct15:01
d34dh0r53#topic roll call15:01
gtemao/15:01
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, deydra15:02
d34dh0r53hey gtema 15:02
gtemahey hey. I was in the train to FOSDEM on friday and not sure you got my reply15:02
* d34dh0r53 is working on a bespoke artisan ping for dmendiza 15:02
d34dh0r53gtema: yeah, I saw it, no worries15:03
d34dh0r53#topic review past meeting work items15:05
gtemaare we alone?15:05
d34dh0r53#link https://meetings.opendev.org/meetings/keystone/2026/keystone.2026-01-14-15.04.html15:05
d34dh0r53Yeah, I think so15:06
d34dh0r53no action items from last week15:06
gtemauhm, makes no sense to make a meeting in full form I guess15:06
dmendiza[m]lmao, appreciate the artisan ping15:06
d34dh0r53lol15:06
gtemaoh nice :)15:07
d34dh0r53#topic liaison updates15:07
d34dh0r53just going to continue quickly for the logs since I started the meeting15:08
d34dh0r53nothing from me15:08
gtemanothing on my side either15:08
d34dh0r53#topic specification Secure RBAC (dmendiza)15:08
d34dh0r53#link https://governance.openstack.org/tc/goals/selected/consistent-and-secure-rbac.html#z-release-timeline_15:08
d34dh0r532026.1 Release Timeline15:08
d34dh0r53Update oslo.policy in keystone to enforce_new_defaults=True15:08
d34dh0r53Update oslo.policy in keystone to enforce_scope=True15:08
dmendiza[m]Yeahhhhh,  I haven't done much upstreaming lately 😅15:09
dmendiza[m]New year's resolution not going great.15:09
gtemaI'm so sorry for you15:09
d34dh0r53no worries, thanks dmendiza 15:11
d34dh0r53#topic specification Secuirty Compliance Testing (dmendiza)15:11
d34dh0r53#link https://review.opendev.org/c/openstack/devstack/+/95796915:11
dmendiza[m]Same same15:12
dmendiza[m]I need to bring back Upstream Fridays15:12
d34dh0r53👍15:13
d34dh0r53#topic specification OpenAPI support (gtema)15:13
d34dh0r53#link https://review.opendev.org/q/topic:%22openapi%22+project:openstack/keystone15:13
gtemano changes on that particular side15:14
gtemaMaybe I should drop it from the agenda since amount of changes is very low recently15:14
d34dh0r53ack, how much is there left to do?15:15
gtemawell, I was never starting to work on the oauth2 side and some other of the very esoteric parts15:15
gtemafrom the basic functionality pretty much everything is covered15:16
gtemaSadly I needed to rollback OpenAPI rendering for the docs since python ecosystem broke again15:16
d34dh0r53yeah, that's a bummer15:18
d34dh0r53I'll take it off of the weekly agenda, we can re-add later if needed15:18
gtemamaybe I find time to fix it again, but lets see.15:18
gtemaOk, will remove it for now15:18
d34dh0r53#topic keystone-rs15:19
d34dh0r53#link https://github.com/openstack-experimental/keystone15:19
gtemathat one is big :)15:19
gtema1) I started working on adding service accounts (basically a nonlocal user)15:19
gtemait required to add a new user_option to be able to clearly identify based on the datamodel that we have now15:20
gtema2. at FOSDEM had a very fruitful discussion with jpevrard and got new crazy idea how we can improve service-to-service communication based on the mTLS (but in a manageable way)15:21
gtemaso that the same mechanism can be also used by the users15:21
gtemaThis marks for me OpenStack landing in the zero-trust era fully15:22
gtemaand the same method can be used for the user workload authentication (VM started by nova can authenticate to keystone)15:22
gtemaI am currently trying to write down all thought on that drawing pictures so that we can discuss it in a wider round15:22
gtema3. preparing the kubernetes token review exchange, but that is technically even obsolete by the previous thing. Still I make sense to implement it as well15:23
gtemaso far that's it15:24
d34dh0r53wow, awesome15:24
gtemaour discussion from recently - how services communicate with keystone (passwords, appcreds, scopes, etc) are all covered by this idea15:25
gtemaand it is also a prerequisite for implementing better confidential VMs on OpenStack15:25
d34dh0r53s/passwords,appcreds,scopes,etc/mtls/g or is it a different layer?15:26
gtemathis all belongs together15:27
cardoeKubernetes Token Review would be a +1 from me. We run our OpenStack in k8s. And would love if each service used its k8s service account.15:27
d34dh0r53Yeah, that's also a +1 from us15:27
gtemaI mean that use of mtls helps us to solve the problem of hardcoding credentials between services and struggles of using appcreds due to need of system scope15:27
d34dh0r53Yeah, I wonder how much work it will be for the service accounts to implement mtls15:28
d34dh0r53err, services15:28
gtemalemme bring everything to the e-paper and we can discuss it.15:28
cardoeWe're using trust-manager from cert-manager for doing the CA and TLS certs right now afaik.15:29
cardoeI've gotta jump into a video call in 1 minute but I just wanted to drop an issue I've seen recently.15:29
gtemawe support mtls already now, but it is so unflexible15:29
cardoehttps://launchpad.net/bugs/2139467 and https://review.opendev.org/c/openstack/keystone/+/975342 is the patch (I need to fix a pep8 issue)15:29
gtemaack, cardoe, will look at it15:30
opendevreviewDoug Goldstein proposed openstack/keystone master: avoid crash with federated tokens and fernet  https://review.opendev.org/c/openstack/keystone/+/97534215:30
cardoemaster skyline causes keystone to crash when using any federation15:31
gtema:)15:32
d34dh0r53ack, let's move on to bug review then since we bled into15:33
d34dh0r53#topic open discussion15:33
d34dh0r53#topic bug review15:33
d34dh0r53#link https://bugs.launchpad.net/keystone/?orderby=-id&start=015:33
d34dh0r53several new keystone bugs, starting from the top which is the one we just discussed15:33
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/213946715:34
d34dh0r53we'll review ASAP cardoe 15:34
d34dh0r53next up15:34
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/213923515:34
d34dh0r53looks like we need to clean up some eventlet code in python-keystoneclient15:35
gtemaI've seen we use keystoneclient in the keystonemiddleware. It does not look good from it's deprecation pov and we would been to address this as well15:37
d34dh0r53Yeah, we really need to get on with deprecating it15:37
d34dh0r53next up15:39
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/213872515:39
d34dh0r53This is something we should fix, not necessarily backport though15:41
d34dh0r53next up15:41
d34dh0r53#link https://bugs.launchpad.net/keystone/+bug/213854515:41
d34dh0r53Yep, we need to fix this as well15:42
d34dh0r53Lot's of good keystone bugs for upstream Friday ;-)15:43
d34dh0r53#link https://bugs.launchpad.net/python-keystoneclient/?orderby=-id&start=015:43
gtema:)15:43
d34dh0r53no bugs in keystoneclient other than what we just talked about15:43
d34dh0r53#link https://bugs.launchpad.net/keystoneauth/+bugs?orderby=-id&start=015:43
d34dh0r53keystoneauth is good to go15:44
d34dh0r53#link https://bugs.launchpad.net/keystonemiddleware/+bugs?orderby=-id&start=015:44
d34dh0r53one new bug in keystonemiddleware15:44
d34dh0r53#link https://bugs.launchpad.net/keystonemiddleware/+bug/213852815:45
d34dh0r53we need to kill this code15:45
gtemayeah, right. We need to kill a lot of code ;-)15:45
d34dh0r53indeed15:45
d34dh0r53next up15:46
d34dh0r53#link https://bugs.launchpad.net/pycadf/+bugs?orderby=-id&start=015:46
d34dh0r53nothing new in pycadf15:46
d34dh0r53#link https://bugs.launchpad.net/ldappool/+bugs?orderby=-id&start=015:46
d34dh0r53nor in ldappool15:46
d34dh0r53#topic conclusion15:46
d34dh0r53Thanks folks!15:47
gtemacool. This friday I am here for the reviewathon15:47
d34dh0r53cool, see you then15:47
d34dh0r53#endmeeting15:47
opendevmeetMeeting ended Wed Feb  4 15:47:46 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:47
opendevmeetMinutes:        https://meetings.opendev.org/meetings/keystone/2026/keystone.2026-02-04-15.01.html15:47
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/keystone/2026/keystone.2026-02-04-15.01.txt15:47
opendevmeetLog:            https://meetings.opendev.org/meetings/keystone/2026/keystone.2026-02-04-15.01.log.html15:47
gtemasee ya guys15:48
opendevreviewDavid Wilde proposed openstack/python-keystoneclient master: Remove eventlet support from cms module  https://review.opendev.org/c/openstack/python-keystoneclient/+/97565015:59
opendevreviewGrzegorz Grasza proposed openstack/keystonemiddleware master: Remove ec2_token and s3_token middleware  https://review.opendev.org/c/openstack/keystonemiddleware/+/97529316:03
*** croeland1 is now known as croelandt16:15
andrewbogott__gtema, d34dh0r53, can you assist me with https://bugs.launchpad.net/keystone/+bug/2140354?  I imagine I have something going on in my ldap schema that you didn't anticipate.17:08
gtema:)17:08
gtemaactually enabled is a must. I would rather say it is not a schema issue but your ldap configuration. I barely remember there was some other issue with the enabled flag in ldap17:10
gtemamaybe there is a misconfiguration on your side that went unnoticed till now17:11
andrewbogott__Typically I would consider a new schema requirement that breaks an existing deployment to be a regression, do you disagree?17:11
gtemastrongly - when a bug is fixed it is not a regression17:12
andrewbogott__ok... what bug is now fixed?17:12
gtemaI don't mean a certain bug fix in this particular case - it is a general statment17:12
gtemahttps://docs.openstack.org/api-ref/identity/v3/index.html#show-user-details says the 'enabled' field is not optional and must be returned. Now schema enforces this17:13
gtemaI am not an ldap expert, but do you set the [ldap].user_enabled_attribute and/or [ldap]user_enabled_default in the config? Maybe this is missing? The point is that 'enabled' is far more than a simply stupid parameter, half of the security protections in keystone are relying on it17:16
andrewbogott__the code in keystone/identity/backends/ldap/core.py has an interesting if/elif section which, if neither condition is traversed, fails to set 'enabled' on the user object.17:17
andrewbogott__https://review.opendev.org/c/openstack/keystone/+/958205/5/keystone/identity/backends/ldap/core.py17:17
andrewbogott__so it seems we need a fix there to cover that case17:17
andrewbogott__(that case being the one where all related keystone settings are left as default)17:18
gtemadefaults are never good. Imagine you set the user to disabled in the ldap, but keystone does not see it properly (i.e. named differently) - the user will be eventually still able to continue using keystone17:18
andrewbogott__(which seems like a kind of reasonable case)17:18
andrewbogott__well -- even so this breakage is an extremely weird place. If we're going to reject ambiguous ldap records it needs to be on loading.17:18
gtemacan you check whether https://review.opendev.org/c/openstack/keystone/+/972077 is affecting you17:20
gtemaor https://review.opendev.org/c/openstack/keystone/+/95820517:20
andrewbogott__that second one is the same patch i just pasted above :)  But yes, let me add some debug lines and see what we're actually getting from the ldap server.17:21
gtemaah ok, sorry17:22
andrewbogott__yep, as I suspected, kind is never 'enabled'. I take that to mean that my ldap records simply don't have an 'enabled' setting.17:28
gtemaperhaps17:29
andrewbogott__gtema: I'm having a moment of disorientation. I'm trying to dig through the code that handles the ldap config and the mapping of the 'enabled' flag and I can't find any code in the keystone tree that checks user_enabled_attribute. Is it possible that that was simply never implemented? Or am I missing some oblique uses of that flag?18:41
gtemaas said, I am not an ldap user and therefore this code is also not very famillar with the code. @xek should know better, but he is most likely done for today18:43
andrewbogott__fair18:46

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