Monday, 2026-04-27

opendevreviewVictor proposed openstack/ironic master: Truncate node history event to avoid DB column overflow  https://review.opendev.org/c/openstack/ironic/+/98139905:52
opendevreviewKonstantinos Karampogias proposed openstack/ironic-specs master: Firmware endpoint hardware identity  https://review.opendev.org/c/openstack/ironic-specs/+/98580008:09
*** alegacy_ is now known as alegacy11:37
dtantsurJayF: good morning, let me know if you're up for touching upon GPU in the inventory today.12:01
opendevreviewcid proposed openstack/ironic stable/2025.2: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636112:08
opendevreviewcid proposed openstack/ironic stable/2025.1: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636212:09
opendevreviewcid proposed openstack/ironic stable/2024.2: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636312:10
dtantsurhey folks, we discussed having blocklist for steps, so here's an RFE: https://bugs.launchpad.net/ironic/+bug/215045613:46
dtantsurit has one open question about runbooks13:46
dtantsuranyone is welcome to take this, I won't have time for it13:46
JayF#startmeeting ironic15:00
opendevmeetMeeting started Mon Apr 27 15:00:21 2026 UTC and is due to finish in 60 minutes.  The chair is JayF. 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
opendevreviewVerification of a change to openstack/ironic master failed: Revert "Soft power off before ejecting virtual media regardless of CONF.inspector.power_off"  https://review.opendev.org/c/openstack/ironic/+/98536215:00
JayFGood morning everyone! Since we concluded our giant week of meetings, we get a meeting on Monday morning as our reward.15:00
JayFI'll wait to see if we get quorum.15:00
* dtantsur waves15:00
mumesan[m]o/15:02
abongaleo/15:02
clifo/15:02
iurygregoryo/15:03
JayF#topic Announcements/Reminders15:03
JayFPlease review patches hashtagged ironic-week-prio and also tag patches ready for review with ironic-week-prio15:04
JayFAlso, we're in the 2026.2 Hibiscus release cycle now. It's R-22. No deadlines in sight.15:04
JayF#topic Working Group Updates15:04
JayFI cleaned up this portion of the agenda; I only left asyncio which I think is the only one with activity still15:04
cido/15:04
JayFis that accurate? VXLan working group no longer needed?15:04
kubajjo/15:05
JayFAlso does someone wanna speak to asyncio15:05
iurygregoryno updates afaik15:05
dtantsurWe'll pick it up eventually, but not right now15:05
iurygregorywe had the spec up from dtantsur 15:05
iurygregoryyeah15:05
JayFSounds about what I expected. Please update the etherpad if we need to pop up any other working groups15:06
JayF#topic Discussion Topics15:06
JayFOne item here, and I added it15:06
JayFwe have our work items spec up for review15:06
JayF#link https://review.opendev.org/c/openstack/ironic-specs/+/98601815:07
JayFUnless there is major objection; I intend on merging this Wednesday morning PDT in order to give me time to record a PTG recap video before flying out. 15:07
JayFPlease review and help get this into place.15:07
JayF#topic Bug Deputy Updates15:08
JayFDid we have a bug deputy?15:08
cidI don't think we did but I'm happy to take it up this week.15:09
cidSee what we missed out on during the PTG week and such15:09
JayFSGTM, thank you cid.15:09
JayF#topic RFE Review15:09
JayFThere was an RFE filed by dtantsur, I tagged it approved based on PTG conversation15:09
JayF#link https://bugs.launchpad.net/ironic/+bug/215045615:09
cidYeah, about the open question on the RFE, 15:10
cidGiven the idea of runbooks being pre-approved, i think we don't necessarily need to exclude them15:11
JayFI think there's more nuance than that15:12
JayFrunbooks that are isolated to use by a single project can be created by aproject manager15:12
cidI was just about to say, I believe there's a world where we might still do that for runbooks, I can't think of any at the moment.15:12
JayFwhereas the blocklist config is a decision made at a systems level15:12
JayFIMO we would filter in all cases15:12
JayFdtantsur: thoughts?15:14
JayFthere's probably value in getting a decision made and into the RFE15:14
dtantsurI really don't know, hence I put it as an open question15:14
JayFI always just think of it like RBAC alignment15:14
cid+++15:14
JayFwith config files being "system admin"15:14
dtantsurI'm kind of buying into the different persona argument, yes15:14
dtantsurblocklist = "not on my cloud!!"15:15
JayFso I would say don't allow a project manager to do something a system admin says no to15:15
cidAlthough, a project level runbooks can only run on nodes owned by that project.15:15
JayFcid: yeah but think about a multitenant case, you may not trust the node.owner project15:15
JayFcid: so then them running, e.g. drac.factory_reset, might disable a security functionality you have enabled15:15
JayFI'm going to put a note in the bug that this is the assumed consensus and to bring it up again if someone wants to implement it differently.15:16
JayF#topic Open Discussion15:17
JayFAny additional items to bring up to the group before we shut down?15:17
cidNone from me at this time. By the way, it makes perfect sense that system-level decisions trickle down.15:19
JayFLast item: Does someone wanna volunteer to chair next week's meeting?15:20
JayFIn lieu of someone else to run the meeting; I will but ideally someone not out half this week will run next weeks meeting :)15:20
JayFAight, have a good week all p/15:21
JayF#endmeeting15:21
opendevmeetMeeting ended Mon Apr 27 15:21:58 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:21
opendevmeetMinutes:        https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-04-27-15.00.html15:21
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-04-27-15.00.txt15:21
opendevmeetLog:            https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-04-27-15.00.log.html15:21
clifo/15:22
cardoeWell dang that was quick.15:23
JayFThree-(awake)-person meetings move fast lol15:24
JayFcid: do you wanna pair on drafting up an OSSN for the IPMI console thing?15:25
cidSure115:26
JayFgive me 5 minutes and I'll spin up a zoom15:26
JayFI'll link it in here in case others wanna join15:26
cidclif, I saw your review on the pickle change. I will address it in a moment15:34
cidAlso, I have a change for the same in the oslo_service and oslo_config repos, did you get a chance to check those out as well?15:35
cidIf those land, we may not need the Ironic patches15:35
JayFhttps://us06web.zoom.us/j/87613520814?pwd=cWHtaaiBRSeUaWeX4N5DIOF7JLgfxy.1 is where cid and I are drafting the OSSN if anyone wants to join15:37
clifcid: no, I didn't realize, I'll go take a look15:37
cidtks15:37
opendevreviewVerification of a change to openstack/ironic master failed: Revert "Soft power off before ejecting virtual media regardless of CONF.inspector.power_off"  https://review.opendev.org/c/openstack/ironic/+/98536215:52
opendevreviewJay Faulkner proposed openstack/ironic unmaintained/2024.1: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98641716:07
cardoeSo I'm looking at neutron policies and ironic policies... I notice that `role:admin` without a project seems to be all that's required to modify system wide resources in neutron while ironic uses system scope=all... do we have any RBAC definition / policy on this? Or is that coming hire up from keystone and the OpenStack TC?16:07
opendevreviewJay Faulkner proposed openstack/ironic unmaintained/2023.1: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98641816:07
cardoeTheJulia out?16:18
TheJuliaYes, wife is having surgery today16:19
dtantsurcardoe: definitely today and possible for the rest of the week16:19
dtantsurah, you're online enough :)16:19
opendevreviewMerged openstack/ironic stable/2026.1: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98623516:19
TheJuliaBut, I’m also sitting in the waiting room16:19
cardoeoh that was today. I forgot.16:19
* TheJulia reads16:19
cardoeTheJulia: good luck16:19
TheJuliacardoe: thanks16:19
TheJuliaSo… There *IS* a spec which lays out ironic’s approach16:20
cardoeI'm also doing crazy things and looking at using OPA so I can compile and validate my policies.16:20
TheJuliaPart of the challenge is we were able to make amazing progress on the original Secure RBAC approach, and then the TC decided to embrace the original bug that started secure rbac work16:20
TheJuliaWhich is wrong, but hey, I wasn’t on the TC at that point16:20
cardoeyeah so help me understand so I can drag the TC to the right place.16:21
TheJuliaBasically the bug was that admin is admin everywhere and can potentially in some projects cross the project scope16:21
TheJuliaAnd, we did system scope which is “interacting with the actual system itself as if your the owner of the system16:21
TheJuliaOr the manager/operator of the cloud. So we got it out the door early on, and kind of went “well, we don’t want to blindly just give admins to all our endpoints, so we kept system scope on a lot of them which were more information or background16:22
TheJuliaLike.. no, any admin shouldn’t be able to get service hostnames16:22
TheJuliaThat is by definition through the exposure of internal hostnames, information leakage.16:23
cardoeYeah. I'm embracing that.16:23
TheJuliaIt’s info only useful for the operator of the infra or maybe their most trusted users who transcend the boundary16:23
cardoeI'm starting to write this "personas" doc.16:23
TheJuliaAnyhow, we shipped system scope, and some operators who are properly scoped loved it and gave super positive feedback so we kept it.16:24
TheJulia++16:24
cardoeAnd so that model maps better to the system scope.16:24
cardoeI'd also like to add an "audit" role.16:24
* TheJulia goes to the cafe on the other side of the hospital lobby where she suspects they put happy meds into the mochas16:25
TheJuliacardoe: What would audit do?16:25
TheJuliaPlease assume that with a twist like Lucy Darling, since I’ve been viewing her videos this morning.16:25
cardoeRead the metadata on objects but not necessarily the objects themselves.16:25
TheJuliaHow does reader not fulfill that need?16:25
cardoeI cannot think of any in ironic that would work with audit.16:25
opendevreviewVerification of a change to openstack/ironic stable/2025.2 failed: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636116:25
cardoeBut I can in nova and barbican.16:25
JayFdtantsur: https://review.opendev.org/c/openstack/ossa/+/986424 please review16:26
TheJuliaOr better yet, how would it fit into ironic’s world and user flows?16:26
TheJuliacardoe: fair enough16:26
JayFironic cores, really, please review https://review.opendev.org/c/openstack/ossa/+/98642416:26
TheJuliacardoe: maybe… node_history ?16:30
cardoemaybe16:30
cardoeWe've got service accounts that globally want to check the connectedness between objects.16:30
cardoeOr to see when some resource was deleted or created.16:31
cardoeJust thinking about the principles of least access.. they don't need to actually read data about the objects but just confirm metadata about them.16:31
TheJuliacardoe: node list would be necessary, our pattern is largely just to redact if you don’t have access but our model is much more “if you got permissions to read, we give you a lot”16:31
TheJuliaWhich makes it no different from reader, although maybe the “all node history” idea/patches16:32
TheJuliaWhich I’m super good with idea wise16:32
* TheJulia sips the mocha as the atoms align16:33
cardoeThis is just an early white board idea.16:33
TheJuliacardoe:  need any additional context?16:33
cardoeSo far I'm just sitting down and thinking "I am needing to do this... what would be job duties entail"... now the same person might have multiple roles.16:34
cardoeI'm just thinking in terms of Gherkin16:35
cardoeThe behavior driven development domain specific language.16:35
TheJuliaI tend to think of audit more being suitable for processes, like system scope “audit” would be amazing for delineation of access rights for a global node history endpoint16:35
cardoeYeah16:35
TheJuliaIt wouldn’t need much else on our api surface in such a case.16:35
TheJuliaCool, cool. Going to return to my cloaking field in that case.16:38
JayFJFYI, OSSA-2026-008 will be landing and announced in the next hour or two. This is for the IPMI Console injection stuff. CVE is pending but we don't have to wait on that to announce.16:53
JayFSo much easier to do this stuff with the embargo off :)16:53
opendevreviewMerged openstack/ironic stable/2025.1: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636217:05
opendevreviewMerged openstack/ironic stable/2024.2: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636317:06
opendevreviewTakashi Kajinami proposed openstack/ironic master: Remove deprecated http_basic_* options  https://review.opendev.org/c/openstack/ironic/+/98642917:06
opendevreviewcid proposed openstack/ironic master: Bump release mapping to 35.1.0  https://review.opendev.org/c/openstack/ironic/+/98643517:55
cidJayF, release cutting attempt, I have these two changes up now ^^ && https://review.opendev.org/c/openstack/releases/+/98643417:57
cidHow detailed should the commit messages be?17:57
JayFLook at previous messages in that dir :) that's what I'd do17:58
opendevreviewcid proposed openstack/ironic master: Bump release mapping to 35.1.0  https://review.opendev.org/c/openstack/ironic/+/98643518:14
opendevreviewMerged openstack/ironic stable/2025.2: Shell-quote console command passed to `socat`  https://review.opendev.org/c/openstack/ironic/+/98636119:47
guilhermesp___Hey all! — anyone running Debian (bookworm/trixie) cloud images on software-RAID1 + UEFI via Ironic (we're on IPA master since LP#2090993 fixed the "no space left on ESP" issue for us)? Hitting initramfs-without-mdadm and grubx64.efi-without-mdraid1x; works on Ubuntu after custom DIB hooksi took  from https://github.com/osism/openstack-ironic-images but wondering if there's a supported path. Also: on delete,  20:02
guilhermesp___NVRAM entries added ahead of NIC so the node never PXEs back for cleaning without a manual F11 — anyone solved that cleanly? btw its a supermicro box 20:02
JayFguilhermesp___: for the second one, baremetal node set <node> --driver-info force_persistent_boot_device=Always20:08
JayFguilhermesp___: for the first one, I thought we had a patch around that. How old is the ironic you're running?20:08
JayFoh, you were saying you had a fixed version20:11
JayFI misread the first time through20:11
JayF> Hitting initramfs-without-mdadm and grubx64.efi-without-mdraid1x20:11
JayFI have no idea what this means20:11
JayFand I'm unfamiliar with osism20:11
guilhermesp___hey JayF ! thank you for the prompt reply! Yeah i tried the Always option but no luck =( 20:12
guilhermesp___https://www.irccloud.com/pastebin/2i6pDkeB/20:12
guilhermesp___im running 2025.1 but ipa is running master 20:12
guilhermesp___to get LP#2090993... then when i deployed, i got grub rescue right away 20:12
JayFAight, I think this is buggable then. I don't have any quick/easy answers for you :( 20:13
JayFSomeone else might, but I don't. 20:13
JayFguilhermesp___: look at clean_uefi_nvram step, perhaps if added to cleaning at the right spot might help?20:15
guilhermesp___huuum but i guess its kinda chicken-egg problem since its a clean step and it should pxe boot at first place to be able to run? 20:16
JayFoh, because this is *on delete*20:16
JayFugh20:16
JayFthis sounds like a borderline hardware bug?20:16
guilhermesp___thats something im suspecting of yeah... i already deployed nodes with ironic + supermicro but none of them were using software raids, thats the diff here 20:17
guilhermesp___btw, for a ubuntu image, i had to get something like this to make it fully boot ( with a grub menu and all ) https://paste.openstack.org/raw/betYbySCfV3kG8IOHAXI/20:17
guilhermesp___but it sounded like toooo much to customize to make it workable lol 20:18
JayFkubajj and hjensas are often our experts on raid stuff20:18
JayFthey might have some insight I'm missing20:18
JayFguilhermesp___: getting software raid working on Ironic is not trivial, and image work is a large part of it20:18
guilhermesp___yeah yeah agreed, i have a bit of success with alma linux in the past but i still had to do some of this osism recipes baked 20:20
JayFI'd file and bug and see if Kuba or Harold have any more info.20:20
guilhermesp___yeah good call 20:21
kubajjguilhermesp___: at CERN we use RHEL and Alma Linux with RAID1. Both IPA and Ironic are 2025.1. I think we do not have any issues. Can have a look at some of the issues you listed tomorrow 20:34
guilhermesp___yeah right! i had no issues with rhel and alma based not long ago! im trying to write a up a bug into LB ironic here with as much details as possible on the debian family 20:37
guilhermesp___https://bugs.launchpad.net/ironic/+bug/2150502 ok thats more or less what i tried so far  and as we iterate over it, i might add the custom dib elements if that matters 20:48
opendevreviewJay Faulkner proposed openstack/ironic-specs master: Add 2026.2 Work Items document  https://review.opendev.org/c/openstack/ironic-specs/+/98601820:59
JayFIronic cores **please** review this! Nobody has looked at this yet today, only me/jand ers/c.i.d. have comments on it at all. I'm flying out of town on Wednesday and really want this landed by then.21:00
opendevreviewMerged openstack/networking-baremetal master: Remove binding:host_id setting for L2VNI subports  https://review.opendev.org/c/openstack/networking-baremetal/+/98520621:36
opendevreviewMerged openstack/ironic master: Revert "Soft power off before ejecting virtual media regardless of CONF.inspector.power_off"  https://review.opendev.org/c/openstack/ironic/+/98536221:36

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