Monday, 2026-07-06

opendevreviewMerged openstack/ironic master: api: Add schema for portgroups API  https://review.opendev.org/c/openstack/ironic/+/95988109:02
opendevreviewMerged openstack/ironic master: api: Add schema for volume_target/_connector API  https://review.opendev.org/c/openstack/ironic/+/99328409:09
sean-k-mooneyclif: https://review.opendev.org/q/topic:%22startup-time%22 i rebased/updated the startup time patches into one seriese if you want to check them out again09:45
sean-k-mooneythat seriese also neeed https://github.com/openstack/nova/commit/2714c71d05e8501ddb3a8fc0c36f3406ef974072 which is already merged09:48
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569709:51
opendevreviewMerged openstack/python-ironicclient master: Add portgroup uuid to vif attach OSC client  https://review.opendev.org/c/openstack/python-ironicclient/+/99040910:34
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569712:32
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569712:59
TheJuliagood morning13:03
opendevreviewKonstantinos Karampogias proposed openstack/ironic master: Add vendor/model/serial_number to firmware component endpoint  https://review.opendev.org/c/openstack/ironic/+/99203413:06
opendevreviewNuutti Hakala proposed openstack/ironic master: Limit LLDP TLV value length and interfaces  https://review.opendev.org/c/openstack/ironic/+/99045213:30
dtantsurmorning TheJulia 13:36
iurygregorygood morning ironic13:39
TheJuliahttps://github.com/openstack/ironic/pull/14 <-- I thought there was a bot which handled people opening PRs?13:43
iurygregory_I also thought there was a bot that handle that, I do remember seeing something like this...13:56
cardoeThere was and I think someone mentioned it was busted.14:01
JayFFyi, I will not be at the meeting today.14:01
TheJuliak14:02
TheJuliacardoe: figures14:03
cardoeA Racker opened one recently and it didn’t work right.14:06
cardoeSo I kinda let fable go crazy over the weekend and eat all my usage on ironic.14:06
TheJuliaoh. my.14:07
cardoeSo most of it wasn't code. It was analysis which I believe is what Fable is suppose to be better at.14:09
cardoeWell not all my Fable usage. My son has been wanting to learn programming. And we had come up with an idea of writing an iOS game that he could put in the app store for free and send his family a link to it to show off what he did. Given how the world is changing with the LLMs I decided to focus more on the programming concepts and behaviors and let him use the LLM for the actual code but told it to write lots of comments.14:11
cardoeHe saw a pop up that Fable is the best model so lemme use that...14:11
cardoeWish I had a CLI to launchpad cause I'd create a bunch of issues as reminders for the stuff.14:20
JayFthe API is trivial, but I've only used it unauthenticated/ro14:23
opendevreviewDoug Goldstein proposed openstack/ironic master: add ironic-manage to CLI references doc section  https://review.opendev.org/c/openstack/ironic/+/99610414:26
cardoeJayF: if I'd like https://review.opendev.org/c/openstack/ironic/+/989431 backported, do I need to tell the creator to make a release note? Or should I just push one for them?14:27
TheJuliaIts not that complicated to auth against based upon some of the examples I've seen over the years.14:28
JayFcardoe: yes14:28
JayFcardoe: that change needs a release note anyway14:29
* TheJulia is all for doing release notes for folks14:29
JayFcardoe: and it's questionable if that behavior change can be backported14:29
TheJuliaI think its a bugfix, fwiw14:29
TheJuliaJust, one which is semi-rare14:30
JayFyeah, I'm not saying it can't be, I'm saying it's a clear behavioral change14:30
TheJuliamore so because most ironic operators are using elevated credentials14:30
TheJulia++14:30
JayFso it's a matter of how it's framed/what the failure/success cases are14:30
cardoeSo been playing with having some roles with less access or targeted access14:30
JayFif this is relevant context for 989431 make sure it also gets into the commit/release note/bug/etc :D 14:31
cardoeWell I'm more than happy to rewrite the commit message and the release note.14:31
cardoeCause the issue is a bit gross14:31
TheJuliacardoe: I'd say go right ahead14:32
cardoeIf someone does a manual service or cleaning with steps, the network the device gets put on has ports that are created in the user's context which fails if they don't have access to the service/cleaning networks.14:32
cardoeand it kinda throws some lego pieces all around14:32
TheJuliaIts admin set networks, so.. yeah14:33
TheJuliaor at least, *should* be admin-set networks14:33
JayFyeah that should 100% be detailed in the release note then backported14:33
fricklergithub bot is broken, yes. it leaks credentials, thus it is effectively not running currently14:33
cardoeyeah that's why I'm asking if I can write one on behalf14:33
TheJuliafrickler:fun!14:33
TheJuliacardoe: yes, we view that as acceptable behavior14:34
TheJuliacardoe: ideally, you sign-off on the commit as well14:34
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569714:42
TheJuliaso who is running this week's meeting ?14:50
clifsean-k-mooney: ty, I'm going to try to get them whipped into shape very soon, at least the ones we've confirmed downstream14:50
cardoeSo for once I don't have a work meeting over the top.14:51
clifcardoe: why is it a limitation of argparse and not oslo.config itself?14:51
cardoeSo I could do it but I honestly don't know how.14:51
cardoeclif: I think it was to do with old Python limitation and just hasn't been updated for newer Python14:52
clifaha, well good thing we live in the future14:52
clifI'm going to spend some time in-between finishing the kernel params patches and nova startup perf to add type coercion checking to oslo.config14:53
cardoeyeah like I wanted to toss a ton of code in pbr since the behavior was in stdlib from Python 3.8+ and pbr's behavior was slightly different. And I was told they tried to maintain compat with Python 2.614:53
cardoeThat should not be the case with oslo.config14:53
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569714:59
* TheJulia blinks15:00
cardoe#startmeeting ironic15:00
opendevmeetMeeting started Mon Jul  6 15:00:53 2026 UTC and is due to finish in 60 minutes.  The chair is cardoe. Information about MeetBot at http://wiki.debian.org/MeetBot.15:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:00
opendevmeetThe meeting name has been set to 'ironic'15:00
TheJuliacardoe: you beat me to it!15:01
clifo/15:01
TheJuliao/15:01
rpittauo/15:01
dtantsuro/15:01
cardoeTheJulia: consider it my gift to you since you've been on vacation. And I've needed to do it.15:01
Mahnooro/15:01
* TheJulia hands cardoe a fearless leader shirt15:01
cardoeI believe that's enough for a quorum.15:01
cardoe#link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting15:02
cardoe#topic Announcements / Reminders 15:02
cido/15:02
cardoePlease review outstanding changes in the ironic-week-prio topic.15:03
cardoe#link https://tinyurl.com/ironic-weekly-prio-dash15:03
cardoeDoes anyone have any other announcements?15:03
cardoeAlright moving on.15:03
cardoe#topic Working Group Updates (Async IO)15:04
cardoe#link https://etherpad.opendev.org/p/ironic-asyncio15:04
dtantsurI think iurygregory_ is just back from PTO, so likely no updates15:04
cardoeFair. I saw some patches sitting in the queue so I think he'd appreciate some reviews.15:05
cardoe#topic Working Group Updates (Security Coresec team)15:05
cardoeAny security team updates?15:05
dtantsurImagine a very load scream :)15:05
dtantsurloud15:05
TheJuliaEndless, right ?15:05
* dtantsur slowly nods15:06
cardoeSo you want me to bury my Fable analysis?15:06
TheJuliaSo, I don't think there is much of an update this week, but please keep an eye out this week.15:06
dtantsurQueue up, we already have a backlog to work with :)15:06
TheJuliaheh15:06
cardoeAlright. Then we'll move on.15:07
cardoe#topic Bridging the Gap Gazpacho Cycle Survey and Metrics Analysis (fungi) 15:07
TheJuliaWas that the week before last?15:07
cardoePossibly, I was out then and I thought I saw it when I reviewed.15:08
cardoeThe meeting logs15:08
TheJuliaor weeks before that? I just don't remember and it feels like a repeat since I've already had someone express displeasure over how metrics are being modeled/handled/framed15:08
cardoeokay. We can move on.15:08
cardoe#topic Bug Deputy Updates15:08
cardoecid: what ya got for us?15:08
cidJust two bugs this week, 3 the last15:09
cidI triaged one as invalid15:09
cardoeOur software is 33% better in just 1 week!15:09
TheJulialol15:10
cid:D15:10
cidhttps://bugs.launchpad.net/ironic/+bug/2158578 : Graphical console periodic task should detect stopped containers15:10
cidhttps://bugs.launchpad.net/ironic/+bug/2157002 : BIOS setting update never completes with some vendors15:10
cidhttps://bugs.launchpad.net/ironic/+bug/2157027 : ipxe bootfile_name_by_arch causes boot_mode check to be skipped15:10
cidhttps://bugs.launchpad.net/ironic/+bug/2158986 :15:10
cidhttps://bugs.launchpad.net/ironic/+bug/2158976 :15:10
cardoeSo 2158986 is mine.15:10
cardoeI cannot decide if its a documentation issue or something we should change.15:11
cardoeThe bug was looking for opinions.15:11
cardoeEssentially if you set something like "default_boot_interface = ipxe" then EVERY hardware type you put in your config MUST report that it supports "ipxe" otherwise the conductor will not start15:11
TheJuliaThat feels like a lot of issues15:12
cardoeThe documentation states that every hardware type must support at least ONE enabled_*_interface to load15:12
TheJuliaI think 578 makes sense to just approve as an rfe enhancement15:12
cardoeBut default_*_interface says when there's multiple interfaces for the hardware type, this is the one that will be used.15:12
cardoeBut if my hardware type only supports 1 interface, default shouldn't matter from the reading of the docs.15:13
opendevreviewJed Preist proposed openstack/tenks master: Add Redfish support  https://review.opendev.org/c/openstack/tenks/+/99569715:13
TheJulia002 seems like a legit bug15:13
cardoeAgreed on 578 needing to be done as an enhancement15:14
cardoe002 is a legit bug. There's a patch in ironic-week-prio waiting for review.15:14
TheJulialooks like 027 is already fixed in master branch15:14
TheJuliacardoe: I didn't see that one linked15:15
TheJuliaif you could make sure its linked, that would be awesome15:15
cardoeYeah the author didn't link it.15:15
TheJuliafor 986... you've got a solid point15:16
TheJuliaI'm not sure the right path to "fix" it is more configuration15:16
TheJuliasince its largely an operational disconnect due to docs15:16
TheJuliabut we've seen folks lean hard towards the default values in the past and get into similar siutations15:16
TheJuliasituations15:16
opendevreviewDoug Goldstein proposed openstack/ironic stable/2026.1: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611015:16
TheJuliaI guess I'm not 100% sure docs would fix that, but maybe we should also be actively discouraging the use of driver defaults?!15:17
opendevreviewDoug Goldstein proposed openstack/ironic bugfix/37.0: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611115:17
cardoeWell that was my other thought... we deprecate the idea of defaults.15:17
cardoeAnd if ya want something else, you supply it when you create the node or set it after the fact.15:18
TheJuliaYeah, I'm on the fence on that idea15:18
TheJuliabut I think maybe I might need more or less coffee to reasonably think about that15:18
TheJuliadefaults make ironic easier too, but they lead to pitfalls exactly like this15:18
TheJuliaI feel like "more" with "less work" is the documentation/help text revision15:19
TheJuliaI feel like new options are just a lot of more risk because then people totally need to have the mental model in their heads15:19
TheJulia(and we do, but users, not so much)15:19
cardoeIt came up because a number of deployment docs or tooling use "default_deploy_interface = direct"15:20
cardoeThere was confusion about the priority that an interface is selected when I looked at kolla docs for example.15:21
cardoeif you have enable_deploy_interface = ramdisk,direct which one will be used?15:21
TheJuliayeah, its always priority order of availability from first to last15:22
cardoeThe has supported_deploy_interfaces = [direct, ramdisk]15:22
TheJuliaso, if they remove the default, they will end up with ramdisk15:22
TheJuliaThis is the challenge, it was all well documented in the design docs, but that doesn't necessarilly translate to operational how15:22
TheJuliaand the convience of settings take over in that transition15:22
cardoeYep.15:23
cardoeWell I'll take this one on with some documentation improvements.15:23
cardoeBecause ultimately what folks can achieve with this is enable_deploy_interfaces = direct,ramdisk15:23
TheJuliaYeah, I think that is likely the cleanest path, and most likely for folks to notice if we also issue some sort of reminder in release notes15:23
cardoeand they don't need default_deploy_interface15:23
TheJuliayup15:23
TheJuliabut they need to understand, in as few words as possible (i.e. not llm output), how that configuration translates15:24
cardoeYeah. I'll human write it not LLM write it.15:24
TheJulia:)15:24
cid#link https://bugs.launchpad.net/ironic/+bug/2158976 <- any feedback on this one15:24
TheJuliayeah, I'm pondering that one15:25
TheJuliaWe're in the union of security hardening and side effects, combined with they want private images15:25
TheJuliawhen Ironic was largely geared around public cloud/operator approved images15:25
TheJuliaAll *do-able* but yeah....15:25
TheJuliaI feel like they make a valid point about access but that would need to be a whole new feature to dig into the access and translate that across15:26
TheJuliaI think their use case is a little off from the intent which is why they are having the experience they are as well15:26
TheJuliaand yeah15:26
opendevreviewMerged openstack/ironic-specs master: Move standalone-networking to done in 35.0  https://review.opendev.org/c/openstack/ironic-specs/+/99594415:26
opendevreviewMerged openstack/ironic-specs master: Fix the horribly outdated release index  https://review.opendev.org/c/openstack/ironic-specs/+/99594715:26
cidRight! The reporter has a point even though I don't think the improvement on that code path was literally to share all images with the conductor15:26
cidBut that all depends on what their workflow is15:27
cardoeYeah the reporter has a valid point15:27
TheJuliaabsolutely15:27
TheJuliaand yeah, the only other option then is "hey, operator, what is ironic's project ID"15:27
TheJuliaand depending on how they ran their configuration/setup, that could be problematic15:27
opendevreviewDoug Goldstein proposed openstack/ironic stable/2025.2: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611315:27
TheJuliabecause it is all rooted in the credentials and the project for the account used to configure it15:27
cid++15:28
cardoeWe almost need an image model option15:28
cardoe"public" model and "private" model... Bad names15:28
TheJuliayeah15:28
TheJuliaI think we're going to need to ponder this one a little bit, in all honesty15:29
cardoeYeah I'll make a point to read it and the code and come back to it.15:29
cid\o15:29
cidallow_image_access_via_auth_token is still there fwiw. For operators that will make that explicit config, Maybe we keep it beyond the H cycle.15:29
TheJuliaYeah, I'm also the last person to dig into that area AFAIK15:29
TheJuliayeah15:30
TheJuliathat might be the path, in all actuality15:30
TheJuliaIt doesn't "feel right"15:30
TheJuliabut lets think/ponder, and circle back15:30
TheJuliawho wants to take ownership of comms on this one?15:30
cardoeSounds good.15:30
cid+++ Noted. 15:30
cidFor the time being, I will just lift the invalid tag15:31
cidon the bug for now15:31
cardoeI'll set a reminder to ping the 3 of us and we can decide from there.15:31
cardoeWho will do comms and such.15:31
cardoeWho wants to be our next bug deputy?15:31
MahnoorI can do it this week15:32
cardoesounds good.15:32
cardoeMoving on.. No RFEs in the list.15:32
cardoeThere are some in the ironic-week-prio15:32
cardoe#topic SRBAC enforce_scope removal: always enforce scope in RBAC15:33
cardoe#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/M72AY5ABQFXQ7XHLVEGHLBBK4XFQGVFK/15:33
cardoe#link https://review.opendev.org/c/openstack/oslo.policy/+/98647515:33
cardoeIt's gmaan's topic but he said he probably wouldn't be able to attend15:34
cardoeHere are his notes15:34
cardoeAs ironic support system scope policy, I would like the team to check the impact of enforce_scope removal, and if it is ok to make scope enforcement always enabled and not configurable.15:34
cidSome of those were carried over from the previous week's agenda.15:34
cardoeIf the ironic team thinks it should remain configurable, we can add a flag (something like enforce_ironic_scope, if it does not exist) on the ironic side to control it instead of oslo.policy? 15:34
cardoeah well did we have resolution?15:34
cidNope. The meeting was deferred to today15:35
TheJuliaso, I think the right thing to do is just to remove the option and not worry about configurability, but I don't have operator insight if they have been disabling the option15:36
cardoeYeah15:36
cardoeI unfortunately think what Ironic does is correct and the other projects are wrong.15:36
TheJuliadownside, that is going to get interpretted as breaking for folks who *have* been disabling it, but they were warned a very very long time ago15:36
TheJuliayeah15:36
cardoee.g. global scoped neutron resources should be controlled by system-scope access15:37
TheJuliaYeah, we rolled forward on original scoped design15:37
cardoeNot by any project scoped admin15:37
cardoeI also think it's just removed and don't worry about configurability.15:38
opendevreviewDoug Goldstein proposed openstack/ironic stable/2025.1: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611515:39
cardoeSilence means move on?15:39
cardoe#topic Open Discussion15:39
TheJuliaThen I think that is the right path to proceed along unless someones screams soon15:39
cardoeWhat do folks have for open discussion?15:39
TheJuliavacation was sort of relaxing15:39
TheJuliaAlthough, I chose a hotel not far from the public viewing screens for the world cup15:39
TheJuliatranslate that to: didn't sleep the best. :(15:40
cardoeWell that's a bummer. Hopefully back to your own bed is a plus.15:40
cid:D. Welcome back15:40
TheJuliaDownside, lots of fireworks due to a particular holiday15:41
TheJulia:)15:41
cardoeI've written a basic "ironic-manage" CLI which takes subcommands15:42
cardoeI was hoping for folks to express how they felt.15:42
cardoeWe've been gathering some checkers like TBN and inspection_rules and I figured they could be as subcommands of that.15:43
TheJuliaI think that is a great idea15:43
cardoeThe first subcommand is around hardware types (drivers) and their interfaces... showing what's available15:43
cardoeI've been writing a blog post for the ironic site on the side for adding your own hardware type15:44
TheJuliacool cool15:44
TheJuliaso going back to that image access issue reported, I'm thinking it might be feasible to allow access to be granted if the image is *owned* by the requesting project15:45
cardoeThe mental model for me / us has been every piece of hardware we're managing in one of our racks for provisioning that's hooked to the switch will also be a node in Ironic15:45
cardoeSo we've been extending the usage of "fake-hardware" with some other hardware types15:45
TheJuliaThe downside is we're going to have to pass the requestor's project_id *WAY* down into show and download15:45
TheJuliaand do that consistently, unless it is on *context* somehow, but I don't think it gets preserved there15:46
cardoeThis has led to me writing a bunch of typing on a number of our classes and interfaces.15:46
TheJuliasince that is conductor context most likely15:46
cardoeI've put up the low hanging typing into ironic-week-prio15:46
TheJuliaInteresting, would you be up for writing a blog post?15:46
cardoeThere's a couple of places where things disagree and that'll come as later changes.15:46
TheJuliaon that specifically "creative uses of ironic" ?15:46
cardoeYeah that's what my thought was.15:47
cardoeWe can pass project_id way down. We just need to validate all that.15:47
cardoeAnyway, that's the end of my open topic.15:48
cardoeAnyone have anything more?15:48
cardoeGoing once?15:48
cardoeTwice? Three times?15:49
cardoe#endmeeting15:50
opendevmeetMeeting ended Mon Jul  6 15:50:01 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:50
opendevmeetMinutes:        https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-07-06-15.00.html15:50
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-07-06-15.00.txt15:50
opendevmeetLog:            https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-07-06-15.00.log.html15:50
JayFcardoe: TheJulia: (reading backlog): I wonder if default_*_interface needs to be a mapping with keys of "hardware type" and values of "interface"15:51
JayFthe issue is not that we have a default; it's that we express a single default when in reality it's a "per driver" thing in most cases15:51
cardoeYeah that was one of my suggested approaches in the bug as well.15:53
TheJuliaSort of similar to cardoe's "other" idea :)15:53
TheJuliaI guess I'm gunshy of extending the configuration forward because I've seen people stub their toes on default interfaces before15:53
TheJuliabecause there is a wall mounted footgun trigger right there15:53
TheJuliaAt least, that is how it *feels* for me15:54
TheJuliacardoe: fyi, I asked some clarifying questions on https://bugs.launchpad.net/ironic/+bug/215897615:54
iurygregory_sorry, missed the meeting because I went for lunch with folks from the Office, but yeah no updates, just waiting for reviews in the spec16:08
opendevreviewMerged openstack/ironic-prometheus-exporter stable/2025.1: [stable only] Replace pkg_resources with importlib.resources  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/99476818:32
opendevreviewMerged openstack/ironic stable/2025.1: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611518:35
opendevreviewMerged openstack/ironic stable/2026.1: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611018:44
cardoelast one in the ^ https://review.opendev.org/c/openstack/ironic/+/996111 if someone wants to give it a nudge18:49
TheJuliak18:50
opendevreviewMerged openstack/ironic master: add types to driver interfaces base class for steps and task  https://review.opendev.org/c/openstack/ironic/+/99557118:53
opendevreviewMerged openstack/ironic stable/2025.2: Fix bootfile_name_by_arch to consider boot mode  https://review.opendev.org/c/openstack/ironic/+/99611318:54
opendevreviewMerged openstack/ironic master: Add ironic-manage command for driver discoverability  https://review.opendev.org/c/openstack/ironic/+/99584720:40
JayFJust giving y'all a heads up: in two weeks, the week from July 20-28 (Monday through Monday) I'll be offline and not working on OpenStack. If you need Ironic things, just say something loudly in here and I'm sure someone will help :D 20:49
TheJuliasounds like a plan20:51
opendevreviewMerged openstack/ironic master: annotate the base driver/interface methods as stubs  https://review.opendev.org/c/openstack/ironic/+/99557220:52
opendevreviewMerged openstack/networking-generic-switch master: Add NetconfSwitch base class for NETCONF transport  https://review.opendev.org/c/openstack/networking-generic-switch/+/99006022:23
opendevreviewMerged openstack/ironic master: Truncate node history event to avoid DB column overflow  https://review.opendev.org/c/openstack/ironic/+/98139922:43

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!