15:06:29 <TheJulia> #startmeeting ironic 15:06:29 <opendevmeet> Meeting started Mon Sep 22 15:06:29 2025 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:06:29 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:06:29 <opendevmeet> The meeting name has been set to 'ironic' 15:06:31 <TheJulia> o/ 15:06:38 <iurygregory> o/ 15:06:42 <kubajj> o/ 15:06:49 <TheJulia> Lets see if have quorum! 15:07:05 * TheJulia does the quorum gathering interpretive dance 15:07:35 * dtantsur is late 15:07:42 <rpittau> is even later! 15:07:57 <cardoe> o/ 15:08:08 <TheJulia> Okay, I think we have a basic quorum! 15:08:41 <TheJulia> Looks like we don't have anything beyond our standing agenda today. 15:08:46 <TheJulia> #link https://wiki.openstack.org/wiki/Meetings/Ironic 15:08:56 <TheJulia> #topic Announcements/Reminders 15:09:42 <TheJulia> A general reminder, please take time to review items labeled with "ironic-week-prio". A link for the dashboard is at https://tinyurl.com/ironic-weekly-prio-dash. 15:09:50 <cardoe> The patch queue is pretty big (my metric is 25+) and there's a number of patches with a few +2's that are asking for feedback from other members. 15:10:28 <cid> o/ 15:10:34 * iurygregory promises he will check things today, last week was idrac10 headache... 15:10:37 <rpittau> yeah, I'll check them this week 15:10:37 <rpittau> need to be ready for release 15:10:42 <abongale> 0/ 15:10:54 <TheJulia> This week R-1, so final RC and intermediary releases. Since we've already done our branching and release cuts, odds are we'll get blocked from a .z version update until after the release. 15:11:42 <TheJulia> Anything else for announcements or reminders? 15:12:31 <alegacy> o/ 15:12:46 <TheJulia> #topic Working Group Updates 15:12:58 <TheJulia> First update, where are things at with Standalone networking? 15:13:13 <alegacy> Good progress on the Bifrost+OVS front this past week 15:13:22 <TheJulia> #link https://etherpad.opendev.org/p/ironic-standalone-networking 15:13:49 <alegacy> I was able to get that setup to enroll, inspect, deploy a node using OVS as the switch and separate vlans for inspection, cleaning and tenant networks 15:14:01 <alegacy> ...then delete + repeat to make sure things were still sane. 15:14:02 <TheJulia> That sounds pretty good! 15:14:21 <dtantsur> alegacy: wanna mention the fast-track issue? 15:14:24 <alegacy> I've got the OVS commit almost ready to go. Just need to add a bit of documentation and then i'll present it for review 15:14:42 <alegacy> Yes, dtantsur... thank you for reminding me 15:15:18 <alegacy> So, discovered that if fast-track is enabled it causes issues because we switch the vlan and the node doesn't get rebooted before the next action 15:15:36 <alegacy> that means that it (potentially) has IP address/route info that doesn't align with the current configured vlan. 15:16:22 <alegacy> so short of creating some logic to determine if the network has actually changed before assessing whether fast-track is still viable we short of have to disable it when this feature is on 15:16:40 <alegacy> unless we know some networks will be combined onto the same vlan 15:16:42 <TheJulia> So, technically I suspect there needs to be a hint like line carrier needs to be torn down 15:16:52 <TheJulia> to force re-address collection logic of the OS 15:16:56 <TheJulia> That is, if it is applicable 15:17:05 <dtantsur> I think we should disable it for the first stage, but we need to have a logic of "are we going to change anything" somewhere between the network and the boot interfaces 15:17:44 <dtantsur> in case of Metal3, losing fast-track will annoy a lot of people (including your own colleagues alegacy :D) 15:17:50 <TheJulia> maybe to rephrase into what I'm suspecting, might be good for the boot interface, related to fast track, should likely go "oh, no, can't fasttrack anymore here" 15:17:54 <alegacy> yes, i know ;) 15:18:11 <TheJulia> But may still be navigatable with link state cycling 15:18:15 <dtantsur> TheJulia: yeah, but we need the network interface to tell us that 15:18:53 <dtantsur> Which is tricky in the current design, where the tear_down_ramdisk call does not know why it was called and what is coming next. 15:18:53 <TheJulia> yes, and I think in this state we likely can't just change the vlan, but need to shutdown the port and re-activate the port 15:19:09 <TheJulia> This may also be a larger discussion than this meeting, fwiw. 15:19:17 <dtantsur> Yeah, true. PTG? 15:19:21 <alegacy> the driver currently doesn't take that kind of action. It can, but we'll have to look into that. 15:19:23 <TheJulia> Possibly, yeah 15:19:38 <dtantsur> alegacy: are you planning to join us for the PTG in late October? 15:19:43 <alegacy> yes 15:19:48 <alegacy> i'll make a note of this 15:19:48 <TheJulia> Because there is an amount of context we need to uplevel the discussion to to meaningfully have it 15:20:11 <TheJulia> PTG wise! 15:20:13 <TheJulia> #link https://etherpad.opendev.org/p/ironic-ptg-2026.1 15:20:27 <alegacy> thanks TheJulia 15:20:53 <TheJulia> Done with standalone networking for now? 15:21:04 <TheJulia> i.e. are we good to proceed? 15:21:04 <alegacy> a couple more comments, if I may 15:21:10 <TheJulia> Sure 15:21:44 <alegacy> I've got some reviews open to some pre-requisite items. I've had some reviews from JayF but would appreciate some more eyes on them. 15:21:58 <alegacy> (including a revision of the spec) 15:22:13 <TheJulia> Are they labeled with ironic-week-prio ? 15:22:17 <alegacy> yes 15:22:26 <TheJulia> okay, thanks! 15:22:35 <alegacy> they were anyway... i'll double-check 15:23:05 <alegacy> last item... I have a dependency between the ngs repo and the ironic repo since ngs now needs a base class from Ironic for the driver. 15:23:46 <alegacy> what's the best way to deal with that? Should I just open the ngs review as a WIP even though the dependency won't be satisfied until my future Ironic change goes in? 15:24:06 <TheJulia> alegacy: could you point us to that specifically ? 15:24:40 <alegacy> i haven't uploaded it... i could expedite that 15:24:52 <dtantsur> It souns unusual to me, won't the driver leave in Ironic proper? 15:25:07 <alegacy> sorry... 2 different drivers here 15:25:13 <TheJulia> It is sort of weird which is why I'm wanting to sort of see it. 15:25:21 <alegacy> the Ironic network driver lives in Ironic, yes 15:25:28 <TheJulia> yes 15:25:36 <alegacy> but the switch driver that the standalone networking service will use will be ngs 15:26:01 <alegacy> so i've added a new entrypoint to it... but it needs to conform to an interface that the standalone networking service owns/understands. 15:26:13 <alegacy> so it creates a runtime dependency from ngs -> ironic 15:26:39 <TheJulia> That seems sort of problematic to do that cross invocation, the call itself should ideally have what is needed 15:26:44 <dtantsur> mmmmm 15:26:56 <alegacy> i'll open a WIP review and point you to what I've done. Maybe there's a better way. 15:26:57 <dtantsur> so it's a driver of the future standalone service? 15:26:57 <TheJulia> I guess, that sort of goes back to could we see what you have now to try and wrap our heads around it 15:27:16 <TheJulia> Yeah, that is the best path, I think. WIP it and then it is easier to have a focused discussion 15:27:23 <alegacy> k, super 15:27:29 <alegacy> that's it from me. 15:27:36 <TheJulia> Okay, regarding eventlet! 15:27:53 <TheJulia> We have what appears to be one outstanding patch: https://review.opendev.org/c/openstack/ironic/+/960209 15:28:03 <TheJulia> Which we'll want to backport since its more about config loading/parsing 15:28:11 <TheJulia> in a specific launch use case 15:28:43 <TheJulia> If there is anything else anyone is aware of, please spread awareness. 15:28:52 <TheJulia> Anything else? 15:29:35 <TheJulia> Since we have no pre-set discussion topics, we will proceed to Bug Deputy Updates 15:29:38 <TheJulia> #topic Bug Deputy Updates 15:29:49 <TheJulia> cid: looks like you updated the agenda to indicate 2 new bugs! 15:29:55 <cid> Yeah, I was about to say, except bug deputy updates 15:30:05 <TheJulia> oh, okay! 15:30:05 <cid> It was a quiet week. There were two new bugs. 15:30:07 <cid> With one that uncertain wether to triage as an RFE or not; https://bugs.launchpad.net/networking-generic-switch/+bug/2125229 15:31:27 <TheJulia> Sort of half and half I think 15:31:47 <TheJulia> Mike (the filer) operates at a scale and might have hit something we don't see normally. 15:32:24 <cid> I guess, it's mostly likely a bug. 15:32:34 <cid> That's all for bug deputy updates. 15:33:08 <cid> I'm always here as a fallback to take it up the next week(s) unless anybody else wants to give it a go. 15:33:57 <TheJulia> I can likely do it this week 15:34:07 <opendevreview> Merged openstack/ironic master: Documentation: image_checksum is ignored for OCI URL's https://review.opendev.org/c/openstack/ironic/+/961672 15:35:26 <cid> \o/, thanks, TheJulia. 15:35:59 <cardoe> dtantsur: https://review.opendev.org/c/openstack/ironic/+/959087 is one that folks are waiting on you 15:36:08 <TheJulia> Since we have no RFE review items, we can move forward to Open Discussion 15:36:12 <TheJulia> #topic Open Discussion 15:36:24 <TheJulia> I guess the only real item I have for open discussion is to just highlight the PTG etherpad 15:36:31 <TheJulia> https://etherpad.opendev.org/p/ironic-ptg-2026.1 15:37:09 <iurygregory> I will take a closer look at the topics, we haven't booked the slots I will take this action 15:37:32 <TheJulia> We *should* also likely begin trying to scope the size of the discussions 15:37:44 <iurygregory> ++ 15:38:16 <iurygregory> Do we have any topics that will require cross-team discussions? 15:38:56 <iurygregory> I will send an email today to openstack discuss, in case other teams have topics they want to discuss with us 15:39:46 <TheJulia> I think there should likely be a discussion with Nova regarding getting graphical console access for baremetal wrapped up. And with Neutron regarding supporting vxlan based terminations. I expect that whole vxlan topic to be a long-ish discussion where folks like Cardoe and myself need to get on the same page with neutron folks and ensure we have stuff written down or at least taksed to be written so the entire plan is 15:39:46 <TheJulia> clear 15:40:07 <cardoe> +1 15:40:17 <iurygregory> tks TheJulia 15:41:09 <TheJulia> Anything else? 15:41:56 <iurygregory> I would like to get feedback on https://review.opendev.org/c/openstack/ironic/+/961989 15:42:16 <iurygregory> this helped me to get idrac10 provisioned, finally 15:43:22 <iurygregory> in case others have better idea on how we should approach this problem, tl;dr idrac10 requires us to use Slot1 and all the slots in the machine I have shows CD/DVD as mediatypes, but slot2 is returned first =) 15:43:57 <TheJulia> I'd sort of really prefer the driver to have knowledge of that isolated out to the driver, but yeah 15:44:17 <dtantsur> Similarly, I'd really prefer a design where this logic is in the idrac driver itself 15:44:51 <iurygregory> humm ok, I will try to check that and see how to move forward o/ 15:44:52 <dtantsur> maybe a private method on Redfish boot interface that returns acceptable ID's for a System? 15:45:04 <TheJulia> I think I need to think about it and also look at the current state of the code, since I know idrac10 firmware has also highlighted some other logic changes/handling which is needed 15:45:09 <dtantsur> so that the base class returns None (I don't care), where the iDRAC one returns "1" if the system is iDRAC 10 15:45:41 <TheJulia> dtantsur: That was sort of what my brain was noodling but I want to look at the enitre state of things first 15:45:42 <dtantsur> (then at some point we get quirks, and hopefully this logic will become automated..) 15:46:12 <iurygregory> ok, I will investigate this path after lunch =) 15:46:19 <iurygregory> tks dtantsur TheJulia o/ 15:46:34 <TheJulia> cool cool 15:46:37 <TheJulia> What else! 15:46:55 <iurygregory> nothing else from me 15:48:51 <TheJulia> Okay cool, well then if there is nothing else, we can close this meeting and go back to Taking Over The World^W^H^W^H^W... success!? 15:49:11 <iurygregory> take over idrac10 15:49:14 <iurygregory> :D 15:49:18 <dtantsur> \o/ 15:49:45 <TheJulia> cool cool, thanks everyone! 15:49:56 <TheJulia> #endmeeting