Monday, 2026-02-09

opendevreviewJad Haj Yahya proposed openstack/ironic master: Fix Redfish boot mode handling for read-only BMCs  https://review.opendev.org/c/openstack/ironic/+/97552106:33
TheJuliacardoe: oh, agree, but more so on my end trying to extract base requirements and if we can or understand a need which is different, then that is also a data point… and as with all things if enough  data points, then why not at some point to make it more generally applicable06:37
TheJuliacardoe:I forgot about about the nested socket, doh.06:37
TheJuliaI’ll try and sort that today06:38
TheJuliacardoe: 767-400 was the evil aircraft :)06:46
TheJuliaNewer-ish long haul info-tainment systems with older style USB A ports :\06:47
TheJuliaGranted, 1A ports at like… 140 passengers means the odds of infotainment loads really spiking unexpectedly is sort of low.06:49
TheJuliaiurygregory: janders: I don’t know if either of you know Jad, but a release note would be good :)07:30
rpittaugood morning ironic! o/07:43
opendevreviewMerged openstack/ironic master: Fix retrieval of inventory data from Swift  https://review.opendev.org/c/openstack/ironic/+/97582408:08
opendevreviewDavid Nwosu proposed openstack/ironic master: Fix firmware update error handling and test empty MessageId  https://review.opendev.org/c/openstack/ironic/+/97430708:38
opendevreviewJulia Kreger proposed openstack/networking-baremetal master: l2vni baremetal mech driver  https://review.opendev.org/c/openstack/networking-baremetal/+/97388909:25
TheJuliagood morning rpittau 09:29
rpittauhey TheJulia :)09:31
rpittauoh wait, odd hours for you :D09:31
opendevreviewRiccardo Pittau proposed openstack/ironic-prometheus-exporter master: Refactor redfish parser to eliminate code duplication  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/97589109:39
TheJuliarpittau: I'm in Waterford, IE this week09:51
rpittauoh nice! enjoy :)09:56
opendevreviewMerged openstack/ironic master: Add standalone networking service installation guide  https://review.opendev.org/c/openstack/ironic/+/96647110:12
opendevreviewMerged openstack/bifrost stable/2024.2: [stable only][2024.2] Ensure firewalld is unmasked before enabling it  https://review.opendev.org/c/openstack/bifrost/+/97516910:31
opendevreviewJad Haj Yahya proposed openstack/ironic master: Fix Redfish boot mode handling for read-only BMCs  https://review.opendev.org/c/openstack/ironic/+/97552110:40
pas-ha[m]hi team, have a question about redfish-https - is ironic expected to eject boot media when using 'direct' deploy when booting into provisioned instance? reason I am asking is that we have one specific configuration failing in our downstream tests because the non-ejected boot iso that is still seen by provisioned instance, and the config drive supplied by nova - both have 'config-2' label, which confuses cloud-init into not finding11:26
pas-ha[m]the nova one.11:26
pas-ha[m]besides, AFAIU boot iso might have some secrets inside that the provisioned instance not exactly should be able to access... 😕 11:26
iurygregorygood morning ironic11:37
iurygregoryTheJulia, https://review.opendev.org/c/openstack/ironic/+/975521 has a relasenote , or there is another patch ?11:41
TheJuliapas-ha[m]: so ejecting the vmedia, I honestly don't remember, if your referring to the agent token I don't think we would be worried about that.13:52
rpittaupas-ha[m]: in general, yes, Ironic is expected to eject the virtual media, what you're seeing could be a race when transitioning the node into active state, or even a bug, but you need to check the logs to see exactly what's happening, timeline is important here13:52
rpittaualso yes, you're right in the boot iso there are information that the provisioned node should not access, so if the eject is not happening when transitioning to provisioned, we need to understand why13:52
TheJuliawe likely should be ejecting13:52
pas-ha[m]rpittau:  what this boils down13:53
pas-ha[m]to is the following13:53
pas-ha[m]in certain scenarios - at least with redfish-https + ironic node using portgroups - ironic generates boot iso with 'openstack/*' files for networking config, and then the _label function decides that this ISO should have label config-2, and I am really not sure this is right.13:55
pas-ha[m]I'll file a bug + patch and we can discuss there.13:56
TheJuliaThat is right for forcing the device to self configure from vmedia13:56
TheJuliaThe issue, it sounds like is it somehow remaining attached13:56
rpittauconfig-2 is the standard label for the config drive specification13:56
rpittauTheJulia: yeah for some reason the eject is not happening, might be a problem with redfish-https ?13:57
pas-ha[m]so, you mean config-2 on the boot iso is so that cloud-init in it can find "itself"/the same root disk it is booted from, and configure itself.. ok13:57
TheJuliawell13:57
TheJuliaredfish-https should be boot from the iso via the url, nothing else13:57
TheJuliaso there shouldn't be actual attachment13:57
TheJuliapas-ha[m]: yeah13:57
rpittauI mean, if we don't clear the http boot url once the node moves to active state,  the node will find that attached afterwards14:01
pas-ha[m]so does redhish-https boot iface + direct deploy iface has any meaning/is supported?14:01
pas-ha[m]if not, might this be the reason why redfish-https is listed as 'not supporting  config drive'?14:01
rpittaupas-ha[m]: mmm well redfish-https not supporting config drive because of the limitation of the redfish http boot, yeah14:07
pas-ha[m]so what happens with redhish-https + direct, is the final instance boots of boot iso or from kernel/whatever written to disk?14:09
rpittaupas-ha[m]: it hsould boot from local disk14:13
rpittaupas-ha[m]: I suggest you open a bug, I think we may need to just vberify that the url is actually removed before rebooting, but we will need to check logs and code before giving a final answer14:14
rpittaualso please check that there is nothing wrong with the BMC behavior like something blocking the cleanup of the http boot url 14:14
pas-ha[m]BMC is sushy simulator in this particular case14:16
pas-ha[m]Ok, will file a bug14:16
TheJuliaI wonder if the issue here, and this is likely an issue with sushy-tools is the detachment is likley not happening there14:26
TheJuliayeah, so that is exactly it14:26
rpittauwondering the same :/14:26
TheJuliaI thought sushy-tools detaches it, but yeah, its a sushy-tools bug14:28
TheJuliathe base issue and root of the problem with testing http boot funcitonality with vmedia is the e2dk firmware has no concept of forcing it directly so your seeing our single run ci "hack" expecting similarity with hardware14:29
rpittauTheJulia: it's also true that we don't verify the ejection in https://opendev.org/openstack/ironic/src/commit/44184af8135e938b2d44ca83686186373b97c8a1/ironic/drivers/modules/redfish/boot.py#L44914:39
TheJuliathere is no ejection with booting from http14:40
TheJuliait is the reboot14:40
rpittaummm right14:41
TheJuliahttps://opendev.org/openstack/sushy-tools/commit/feb88a5c461955e077151e6c33f1c1327077b13814:41
TheJuliaBasically, sushy tools needs to rip it out upon the next reboot14:44
rpittauyeah14:44
rpittauI can write a patch for that if you're not on it already14:46
Sandzwerg[m]ohai ironic, I have a question. Is anyone aware of ipxe built into the uefi shell and is there a way to use it with ironic?14:46
TheJuliarpittau: that would be much appreciated14:46
TheJuliain team building meetings so can't really work on code stuffs this week14:47
rpittauok :)14:48
rpittaupas-ha[m]: do you have the bug link maybe ?14:53
pas-ha[m]not yet14:54
JayFI can start the meeting in 5 minutes if no one else gets to it before then15:01
JayF#startmeeting Ironic15:05
opendevmeetMeeting started Mon Feb  9 15:05:07 2026 UTC and is due to finish in 60 minutes.  The chair is JayF. Information about MeetBot at http://wiki.debian.org/MeetBot.15:05
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:05
opendevmeetThe meeting name has been set to 'ironic'15:05
iurygregoryo/15:05
JayFHello and welcome to the Ironic meeting. As usual we are operating under the OpenInfra Code of Conduct15:05
janderso/15:05
JayF#note A reminder to all: Please prioritize reviews on our priority dashboard, and feel free to add hashtag "ironic-week-prio" to patches that are ready for review.15:05
cido/15:06
JayF#link https://tinyurl.com/ironic-weekly-prio-dash15:06
JayFWe are in week R-7 of Gazpacho cycle. Only meaningful deadline this week is Extra-ACs deadline15:06
JayFIf you know of any contributors to Ironic who may not have had code commits please contact a DPL asap in order to get them added to the list.15:06
JayF#link https://releases.openstack.org/gazpacho/schedule.html15:06
JayF#topic Working Group Updates15:07
JayFFirst up is standalone networking; alegacy do you have an update?15:07
clifo/15:07
iurygregoryI don't think alegacy is around .-.15:08
JayFWell that's OK, since I think that feature is mostly merged.15:08
JayFWhat about updates for either Async IO or VXLAN Networking.15:08
JayFI believe Julia is out; so I'll plug https://review.opendev.org/q/topic:ironic-vxlan+status:open on her behalf -- code is complete and needs reviews15:09
iurygregoryAsync IO we have the spec under review15:09
rpittauo/15:09
JayFCare to link it Iury?15:10
iurygregory#link https://review.opendev.org/c/openstack/ironic-specs/+/972754/ 15:10
JayFGood stuff. Thanks for that. We're reaching the end of gazpacho so it's a good time to be reviewing things close to the finish line or items for future releases.15:10
iurygregory++15:10
JayFNext up; discussion topics15:11
JayF#topic Discussion Topics15:11
JayFiurygregory: had the only agenda item here15:11
iurygregoryyup15:11
iurygregoryso, I'm working on https://bugs.launchpad.net/ironic/+bug/213736415:12
iurygregorytl;dr HPE machine changes the NetworkAdapter ID after a we do a NIC firmware update :D15:12
iurygregorythis is a bit complicated, because currently we use the ID to know which NIC we are updating prefix `nic:` + NetworkAdapterID15:13
iurygregorythere is a patch up for review 15:13
iurygregory#link https://review.opendev.org/c/openstack/ironic/+/97242115:13
JayFI put a review on that yesterday, the approach looks good but I like Dmitry's ideas around namespacing the id15:13
iurygregorythe only problem I see is if we store that as the `component` on the DB15:14
iurygregorywe will hit the same problem we have in case we update the nic15:14
JayFso you've gotta account for it in parsing code15:15
JayFand it just makes it a thing15:15
iurygregoryoh ok15:15
JayFOK, seems sensible to not wanna go down that path. Can you make sure that's in gerrit15:15
JayFI'm agreeing with you :) wait another 20 seconds before conceding ;) 15:15
iurygregorywe just care about the `nic:<SN>` to identify the right component?15:15
iurygregorywe can go, but it would be a bit hack and ugly I think15:15
iurygregoryunless we have some field that will contain the ID information15:16
JayFthat's a good reason to leave it be15:16
JayFwhich means the namespace isn't a quick or easy win15:16
iurygregoryyeah, at least I have this feeling, not sure what others think 15:16
JayFI'm sold and so far the meeting is me, you, and a few o/'s :D 15:17
iurygregoryrpittau, do you have thoughts on this?15:17
JayFlol15:17
iurygregoryyup!15:17
rpittauiurygregory: I haven't checked the patch, sorry, I'll have a look at it 15:18
iurygregoryrpittau, ack15:18
JayFYeah sounds like that can be worked out in review, let's move on15:18
iurygregoryI will add info in the patch to explain what we discussed here on irc15:18
iurygregoryyup, lets move on15:18
JayF#topic Bug Deputy updates15:18
kubajjo/15:18
JayFI think this is saying iurygregory is the bug deputy? or maybe iurygregory is the next one?15:18
JayFunsure who current one is, but the only reported bug already has a fix merged so15:19
JayFuh, go us?15:19
iurygregoryyeah, I'm the bug deputy for this week15:19
iurygregorycid was last week15:19
cid++15:19
cidThat's right, saw only one new bug for the week15:19
JayFgood work, thanks for being bug deputy next week Iury15:21
JayFThere are no RFEs for review; skipping that section.15:21
JayF#topic Open Discussion15:21
JayFfungi: I assume the Bridging the Gap retrospective is an old item that didn't get cleaned up?15:21
JayFI have an item for Open Discussion: Thursday will be my last day online working on OpenStack through (up to) the end of the month. I'll be out for a minor surgery which will lead to restricted use of my mousin' arm for a couple of weeks.15:22
cidre'Bridging...'. I think that's a left over15:23
clifI'd like ask about this patch I put up to remove shellinabox from Ironic since it's considered abandoned at this point15:23
clif#link https://review.opendev.org/c/openstack/ironic/+/975864/comment/6d06fd5c_0c090d2f/15:23
jandersJayF take care, hope the surgery goes well!15:23
jandersbest wishes15:23
iurygregorytake care JayF o/15:24
JayFclif: I would vote to also remove the iLO console support which depends on shellinabox; the supporting software is old/unsupported enough we should be drastic in our removal.15:24
clifEssentially removing IPMIShellinaboConsole means that the class that IloConsoleInterface inherits from goes away, leaving me wondering what it should derive from now. JayF pointed out that iLO is also going away15:24
fungiJayF: yes, sorry, i should have pulled that off the agenda, was only planning to discuss in-meeting for two weeks, which we did. thanks!15:24
cidJay, best wishes.15:25
cider, JayF.15:25
clifso maybe I should rip out the iLO driver as well in this patch or a direct follow-up patch15:25
JayFthe ilo driver itself can't go until next cycle, with the snmp driver, per deprecation policy15:25
clifJayF: I too hope your surgery goes well and wish you a speedy recovery15:25
JayFhowever I would remove the console support from the ilo driver to enable shellinabox to be fully removed15:25
clifI see, that makes sense15:26
JayFI think about it this way: ideally we would've billboarded loss of the iLO console, too15:26
iurygregory++ to remove the console support like JayF mentioned15:26
JayFbut realistically, nobody should be running this crap in 2026 anyway15:26
JayFand I'd rather rip the bandaid off15:26
cidI had a patch to remove ILo, but I think the review was that there needs to be a deprecation first. https://review.opendev.org/c/openstack/ironic/+/96500915:27
clifwell then I'll leave the driver removal to cid when the time comes15:28
clifbut just remove the ilo console support for now unless someone objects15:28
JayF++ Thank you15:28
cid++, no objections from me. clif, maybe you start the ilo removal with the deprecation patch?15:29
JayFMy belief is we did deprecate it15:29
JayFand if we didn't... whoa boy we're in one15:29
JayFbecaues it's gotta go away next cycle; it hard-deps on the SNMP driver15:29
JayF(proliantutils does)15:29
JayFand proliantutils is near-death15:29
jandersyeah since we're removing iRMC I think it makes sense to try clean up old drivers15:30
janderslooks like the future is converging on Redfish15:30
JayFI guess I'll take an action to see if we formally deprecated iLO, if we didn't I'll do a "formal" patch15:30
JayFbut there's no reality where the SNMP bits of the iLO driver are in the repo after 2026.2 release 15:31
JayFand I don't see a line of volunteers to modernize it15:31
JayFLast call for discussion on this item, or new items for Open Discussion?15:32
jandersjust a quick mention from me15:33
jandersafter a couple months on and off in Europe I am heading back to AUS next week15:33
jandersso unfortunately won't be attending this meeting as much as I have recently15:33
JayFAustralia?15:33
janderswill do my best to keep up being active in reviewing patches though!15:33
jandersyes15:33
jandersit's been great collaborating with you folks in real time for some time!15:34
JayF˙ƃuᴉʇǝǝɯ ǝɥʇ ʇɐ noʎ ssᴉɯ ll,ǝM ¡ɟlǝsɹnoʎ ʎoɾuƎ15:34
jandersJayF LOL! thanks! :) 15:34
clifhold on, gotta turn my monitor upside down15:34
JayFAnything else for the meeting?15:34
JayFThanks folks! Have a good week.15:35
JayF#endmeeting15:35
opendevmeetMeeting ended Mon Feb  9 15:35:46 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:35
opendevmeetMinutes:        https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-02-09-15.05.html15:35
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-02-09-15.05.txt15:35
opendevmeetLog:            https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-02-09-15.05.log.html15:35
kubajjjanders: I was wondering how come you've been around in reasonable times recently (reasonable for me), didn't know you were in Europe15:39
kubajjhope everything goes well, JayF!15:40
kubajjHow bad of an idea is it to change the inspect_interface directly in the DB (question about  https://docs.openstack.org/ironic/2025.1/admin/inspection/migration.html#migration-process step 10). The problem is that I can't change it for active nodes, at least I think15:41
clifJayF: I suspect removing the ilo console stuff is going to break a lot of the ilo driver, just sayin15:53
JayFI am ... extremely skeptical of that statement15:53
JayFespecially given shellinabox is just a frontend to the console15:54
JayFI have time to look at it together if you want?15:54
clifwell I would trust your judgement on the matter, but if I remove IloConsoleInterface then the IloHardware has no support console interfaces left15:54
clifsupported*15:55
JayFYeah, that's 100% OK15:55
JayFconsole is a bonus feature :)15:55
cliflet me poke a little more15:56
clifI just needed to persevere, a lot of tests were breaking but I just needed to remove ilo as a valid console interface16:05
JayFmakes sense16:05
opendevreviewRiccardo Pittau proposed openstack/sushy-tools master: Eject HttpBootUri media on boot target change  https://review.opendev.org/c/openstack/sushy-tools/+/97613816:12
rpittaupas-ha[m]: this should fix the issue ^16:12
rpittauTheJulia: FYI ^16:13
opendevreviewClif Houck proposed openstack/ironic master: Remove IloConsoleInterface since it depends on shellinabox  https://review.opendev.org/c/openstack/ironic/+/97613916:17
iurygregoryJayF, question: do you think we should only use `nic:<SN>` to HPE change it by default and fallback to `nic:<ID>` in case <SN> is not present?16:23
iurygregorycurrently the patch only takes in consideration `nic:<SN>` to HPE, since its where we observed the problem16:24
JayFI think changing things we don't have to change just exposes us to the opposite of this case (Where SN is not valid and ID is)16:24
rpittauiurygregory, JayF, I left a comment on the patch16:54
opendevreviewMerged openstack/networking-baremetal master: Remove MANIFEST.in  https://review.opendev.org/c/openstack/networking-baremetal/+/97598316:56
JayFrpittau: I had already responded to your comment :D 16:58
opendevreviewLuca Del Monte proposed openstack/ironic-prometheus-exporter master: Fix KeyError when timestamp is missing from notification payload  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/97615317:15
opendevreviewLuca Del Monte proposed openstack/ironic-prometheus-exporter master: Fix KeyError when timestamp is missing from notification payload  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/97615318:56
opendevreviewLuca Del Monte proposed openstack/ironic-prometheus-exporter master: Fix KeyError when timestamp is missing from notification payload  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/97615319:01
opendevreviewMerged openstack/bifrost stable/2024.2: Add the ability to override 'disable_boot_timeout'  https://review.opendev.org/c/openstack/bifrost/+/97458819:10
opendevreviewLuca Del Monte proposed openstack/ironic-prometheus-exporter master: Fix KeyError when timestamp is missing from notification payload  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/97615319:53
opendevreviewVerification of a change to openstack/ironic-ui master failed: Remove MANIFEST.in  https://review.opendev.org/c/openstack/ironic-ui/+/97598220:59
opendevreviewVerification of a change to openstack/ironic-tempest-plugin master failed: Remove MANIFEST.in  https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/97599021:39

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