Friday, 2026-01-23

TheJuliacardoe: w/r/t port reconcilliation to ensure switch ports get toggled as it relates to ovn chassis. It sort of looks like, at least at a glance, that your creating a network port on the tenant networking to represent the attachment. wouldn't it be, at least structurally better to keep that out of view of the user but just do enough so the backend binds it. I guess a "fake" network is maybe where my brain is heading into, 00:44
TheJuliawhich doesn't actually do anything but serve as a cross-reference point?00:44
TheJuliaand the record keeper of those attachments00:44
cardoeSo we create a network in the “service” project. The reason is that ports must be attached to a network.01:02
cardoeWe in theory could use that network to allow the network nodes to communicate for HA purposes.01:03
cardoeThe other shortcoming here with the port->trunk->subport model is that there’s no way to indicate if you want that top level port to be the access VLAN or if you don’t want access configured.01:04
opendevreviewSteve Baker proposed openstack/ironic-tempest-plugin master: Add graphical console validate to basic ops scenario  https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/96364301:08
TheJuliaAhh, yeah, that makes more sense. I guess I wasn’t thinking about the network really working . I guess you really can’t say flat me a port and then go “give me a sub port”01:46
TheJuliaBake a network that is independent for controller purposes, then plug the matching vlan on back side. One network per OVN node or chassi. Actually, a network per chassis group simplifies some logic01:49
TheJuliaSo you can do a tight reconcile01:49
TheJuliaTo discuss tomorrow, wife is glaring at me.01:50
cardoeWe've done one per OVN node but I was thinking the same as one network per chassis group as well.02:15
opendevreviewDoug Goldstein proposed openstack/ironic master: feat: skip invalid mac addr interfaces in redfish inspect add speed  https://review.opendev.org/c/openstack/ironic/+/97328302:21
cardoeIf folks would have a look at https://review.opendev.org/q/hashtag:ironic-week-prio+status:open tomorrow. I've tried to curate the list to things that seem ready nor nearly ready with another patch landing.02:23
opendevreviewSteve Baker proposed openstack/ironic master: OciImageService detect bootc image  https://review.opendev.org/c/openstack/ironic/+/96676002:33
opendevreviewSteve Baker proposed openstack/ironic master: Add an autodetect deploy interface  https://review.opendev.org/c/openstack/ironic/+/97318702:33
opendevreviewSteve Baker proposed openstack/ironic master: Replace ironic-tempest-bios-ipmi-direct with -autodetect  https://review.opendev.org/c/openstack/ironic/+/97426002:33
opendevreviewRico Lin proposed openstack/ironic master: Set columns in bios_settings table to bigint  https://review.opendev.org/c/openstack/ironic/+/96834802:40
opendevreviewcid proposed openstack/ironic master: Minimum version bump  https://review.opendev.org/c/openstack/ironic/+/97432107:20
rpittaugood morning ironic! happy friday! o/07:26
jandershey rpittau o/07:55
jandersfolks, if anyone has time I am after second +2 for the monitoring patch: https://review.opendev.org/c/openstack/ironic/+/96694608:26
janderswould be awesome to get it merged before it needs another rebase (I've done ~10 already)08:26
jandersclient and SDK are good to go too: https://review.opendev.org/c/openstack/python-ironicclient/+/967055 https://review.opendev.org/c/openstack/openstacksdk/+/973264. Client had +2 before but also needed rebase / minor changes08:27
jandersCC kubajj JayF stevebaker[m] since they reviewed these previously08:27
jandersthanks and apologies for the noise08:28
rpittaujanders: I'm reading it now :)08:46
jandersrpittau thank you! :)08:46
kubajjjanders: is zuul going to remove its -1 once the change for ironic merges?09:10
kubajj(or do you have to recheck, for example)09:10
janderskubajj just ran recheck09:12
jandersI suspect it had to do with merge conflicts09:12
jandersSDK change also depends on API patch but didn't get the -109:12
kubajjjanders: ok, will finish the review09:13
janderskubajj thank you!09:13
jandersI saw similar behavior with firmware update chain during CI problems - first-in-chain patch passed reviews but was permafailing in CI due to CI issue and other patches in chain got minus votes due to this09:14
jandersdon't think unmerged "base" change should V-1 "depenent" change but I sense it was to do with the merge conflict - if anyone knows for certain I'd be curious to know the mechanism too09:15
jandersno merge conflicts (for now) so reruns should be OK09:15
opendevreviewMerged openstack/ironic master: Add hardware health monitoring via management interface  https://review.opendev.org/c/openstack/ironic/+/96694610:15
janders^ \o/ long time in the making12:22
janders(966946)12:22
cardoekubajj: if you’ve got some time today for some others in https://review.opendev.org/q/hashtag:ironic-week-prio+status:open12:51
kubajjcardoe: will try13:05
rpittaukubajj: metal3 related question (are you on slack btw?), is CERN still using https://github.com/metal3-io/ironic-hardware-inventory-recorder-image ?14:08
TheJuliagood morning14:12
TheJuliabrraaains14:14
kubajjrpittau: was not aware of the slack, but afaik, we do not have a metal3 deployment14:16
rpittaukubajj: yeah, but I remember Arne was using that container outside of metal314:17
rpittauhey TheJulia :)14:17
kubajjrpittau: I will check then14:18
rpittaukubajj: thanks :)14:18
TheJuliarpittau: he did for some add-hoc stuff if I am recalling the last time it was discussed14:45
rpittauok, so no continuous usage? I remember the discussion, but since we plan to archive the repo, I don't want to break them :)14:46
TheJuliaI doub't it14:54
cardoeTheJulia: would you have a gander at https://review.opendev.org/c/openstack/openstacksdk/+/973287 I'm gonna ask the SDK team to land that for the next release so that hopefully we can get the sharding bits completed for NGS15:00
TheJuliareviewed15:01
TheJuliaLGTM, just no workflow button15:01
opendevreviewRiccardo Pittau proposed openstack/ironic master: Migrate database charset from UTF8MB3 to UTF8MB4  https://review.opendev.org/c/openstack/ironic/+/97443015:03
opendevreviewRiccardo Pittau proposed openstack/ironic master: Migrate database charset from UTF8MB3 to UTF8MB4  https://review.opendev.org/c/openstack/ironic/+/97443015:05
opendevreviewTakashi Kajinami proposed openstack/ironic-python-agent-builder master: Drop redundant basepython  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/97443115:07
TheJuliaI'm just going to ask if folks can take some time to look at https://review.opendev.org/c/openstack/ironic-specs/+/959401 it would be much appreciated. We're making really good progress and would love to actually feel like we can proceed with merging things like ml2 plugin capabilities and features15:07
TheJuliarpittau: are you thinking end of cycle for that change?15:13
rpittauTheJulia: yes, I just added a -1 workflow15:13
rpittauput it there so we have time for reviews15:14
TheJuliaack, yeah, that looks pretty good to me15:14
TheJuliawe might want to clarify the release note a little bit more, around the downtime, specificaly it is while each table is locked for the alter table to rewrite the table15:14
TheJuliawhich can take time, quick for smaller operators, larger operators might take a little longer.15:15
rpittauyeah, makes sense15:15
rpittauon a different note, I'm plannig the bugfix cut and while on IPA we're pretty clear, I see we still have some ironic patches open, any one that we'd like to merge before the release?15:17
rpittauI was thinking this for example https://review.opendev.org/c/openstack/ironic/+/97431715:17
opendevreviewRiccardo Pittau proposed openstack/ironic master: Migrate database charset from UTF8MB3 to UTF8MB4  https://review.opendev.org/c/openstack/ironic/+/97443015:21
TheJuliaGood update15:23
rpittau:)15:25
rpittauthanks for the review15:25
TheJuliaso, how do I go about handing out "trust us... merge the vxlan spec" coffee cups?15:26
rpittau:D15:26
TheJuliaI guess I'm excited since we have the base model attachment working15:32
rpittauI'm reading it now before closing for the weekend15:32
TheJuliacool cool15:33
TheJuliaI'm wroking on trying to use up my pre-9am claude credits15:33
TheJuliaSpecifically the agent changes we refer to in that spec15:34
rpittauthat spec is really pushing my limits of networking knowledge to the max15:36
TheJuliaYeah, it... is a LOT15:36
TheJuliabut surprisingly simple once you context load it all15:36
rpittaureally nice reading, I'm a bit concerned about the testing part, but I guess we have external tools besides CI that we can use, and we'll have chance during development to properly test the imeplementation15:43
TheJuliasuper easy and quick review: https://review.opendev.org/c/openstack/networking-generic-switch/+/97352615:55
TheJuliarpittau: fwiw, I've already partially tested the mapping logic, its scarier than it sounds fwiw15:56
TheJuliaerr15:56
TheJuliait sound scarier than it really is15:56
opendevreviewMerged openstack/ironic stable/2025.2: Fix introspection failure handling  https://review.opendev.org/c/openstack/ironic/+/97406916:02
opendevreviewMerged openstack/ironic bugfix/33.0: fix redfish inspect system product name  https://review.opendev.org/c/openstack/ironic/+/97148816:02
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: devstack: Ignore error with file existing on restack  https://review.opendev.org/c/openstack/networking-generic-switch/+/97331016:21
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: WIP: l2vni plug case with Cisco NXOS  https://review.opendev.org/c/openstack/networking-generic-switch/+/96837716:21
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: WIP: Arista EOS and vendor neutral SONiC support for VXLAN attachments  https://review.opendev.org/c/openstack/networking-generic-switch/+/97276316:21
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: WIP: VXLAN: Add Junos, Cumulus NVUE, and denote Dell OS10 as unsupported  https://review.opendev.org/c/openstack/networking-generic-switch/+/97276416:21
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: WIP: OVS testing patch for 'vxlan' binding model  https://review.opendev.org/c/openstack/networking-generic-switch/+/97276516:21
opendevreviewMerged openstack/ironic master: Minimum version bump  https://review.opendev.org/c/openstack/ironic/+/97432116:35
opendevreviewMerged openstack/ironic master: OciImageService detect bootc image  https://review.opendev.org/c/openstack/ironic/+/96676016:43
opendevreviewMerged openstack/ironic master: docs: refactor the inspection docs  https://review.opendev.org/c/openstack/ironic/+/97397116:43
TheJuliathinking outloud, we should likely double check some of our minimum versions and pull them forward16:44
TheJuliagenerally speaking16:44
opendevreviewMerged openstack/ironic master: Refactor firmware update temp field handling  https://review.opendev.org/c/openstack/ironic/+/97431716:55
cardoeYeah pretty sure we won't work with sqlalchemy 0.217:05
TheJuliano, no we will not17:06
TheJulia1.9 at the oldest17:06
cardoeI was making a joke.17:06
TheJuliaoh good17:06
TheJuliabecause I was making stories for the next sprint and hadn't looked yet17:07
TheJuliaooh, 1.417:07
TheJuliathats bad17:07
JayF++17:29
TheJuliaso, 1.4 is actually right, it is where the 2.0 interface was introduced17:43
kubajjrpittau: Arne says, that we used to use it for live-inspection, we have it documented, but I have not used it myself ever17:49
opendevreviewMerged openstack/networking-baremetal master: update pre-commit config to match ironic  https://review.opendev.org/c/openstack/networking-baremetal/+/97397917:51
cardoewell baremetal-l2vni seems to have done the right thing with my hardware17:53
opendevreviewClif Houck proposed openstack/ironic master: Add ConfigLoader class to TBN and use it in TaskManager  https://review.opendev.org/c/openstack/ironic/+/97249517:55
opendevreviewClif Houck proposed openstack/ironic master: Integrate TBN with vif_attach  https://review.opendev.org/c/openstack/ironic/+/97369017:55
opendevreviewClif Houck proposed openstack/ironic master: WIP: Update TBN simulator for vif_attach planning  https://review.opendev.org/c/openstack/ironic/+/97369117:55
opendevreviewClif Houck proposed openstack/ironic master: Add an ordering method for TBN traits  https://review.opendev.org/c/openstack/ironic/+/97451917:55
opendevreviewClif Houck proposed openstack/ironic master: Prevent multiple attach actions being generated for the same port  https://review.opendev.org/c/openstack/ironic/+/97452017:55
opendevreviewClif Houck proposed openstack/ironic master: Filter out already attached portlikes in plan_vif_attach  https://review.opendev.org/c/openstack/ironic/+/97452117:55
clifcardoe: idk if you've tried my patch at all but I found some issues that I've addressed ^^^ you probably want to try from the tip of my branch if you do17:56
cardoeI did run into some issues.17:56
cardoeI've actually got a change for you as well.17:56
clifhaha, sorry :(17:56
cardoeClaude's working through it as we speak17:57
cardoeI'm changing your internal data structures.17:57
cliferg17:57
clifhow so17:57
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline momentarily while we restart it onto new images after an update to the build toolchain18:01
opendevreviewDoug Goldstein proposed openstack/ironic master: convert trait based networking internal classes to dataclasses  https://review.opendev.org/c/openstack/ironic/+/97452418:03
cardoeclif: ^ like that18:03
cardoeSo that's the first step I was going down to fixing what I hit.18:03
clifI like dataclasses, so I generally support this18:04
clifwhat did you hit?18:04
cardoeThe objects are just storing some data fields without specifying what they really need to be. Nor do we limit the inputs in. Somewhere one of the **network or calls like that were resulting in one of the fields having the wrong data.18:05
cardoeSo it could be something I screwed up in my configs but I figured if it was possible for me to load the data wrong we should tighten it up to enforce stuff harder18:06
clifagreed18:07
cardoeWe'll see if that passes tests.18:07
cardoeI also changed tags from a list to a set cause it doesn't make sense to have the same tag repeated.18:07
clifmakes sense18:07
cardoeObviously not an important patch but I'm just trying to make APIs/interfaces where its impossible for the user/developer to use them wrong.18:12
clifwell, if there *is* a bug that this solves it's important enough imo18:12
clifand I was about to +1 it but gerrit is momentarily down18:13
opendevreviewMerged openstack/ironic master: [doc] Releasenote for iDRAC LLDP  https://review.opendev.org/c/openstack/ironic/+/97436618:17
clifcardoe: did you run into anything else?18:33
cardoeWell I didn't get to the underlying issue.18:44
cardoeI'll try your new stuff.18:44
cardoeyeah gerrit being down squirreled me cause I was gonna mention https://review.opendev.org/c/openstack/ironic/+/973557 is another thing I was doing to try and make it hard to use the APIs wrong18:56
opendevreviewMerged openstack/bifrost master: Add the ability to override 'disable_boot_timeout'  https://review.opendev.org/c/openstack/bifrost/+/97391719:13
rm_work[m]JayF: any idea why it would take a long time to show a node within `openstack allocation candidate list` after it goes available in ironic? saw your name on some code around there so maybe you've got some idea what that might be19:24
JayFNo idea at all. That's essentially a nova question moreso than an Ironic one.19:25
rm_work[m]hmm yeah i mean it's placement, but19:25
rm_work[m]wondered if you had worked in that area19:25
rm_work[m]guess not, i will poke nova folks19:25
JayFI've got zero code in placement, pretty close to zero code in nova outside nova.virt.ironic19:25
rm_work[m]hmmm19:25
JayFI tend to lean on John Garbutt for a lot of that stuff, but he's not around IRC so much anymore19:26
*** gmaan is now known as gmaan_afk19:39
TheJuliathe tl;dr is the reconcile loop in progress needs to complete then some amount of time has to pass20:06
TheJuliastictly speaking, there is nothing really stopping us from publishing into placement, but there really hasn't been any push for that to be re-modeled20:08
TheJuliacardoe: I has my testvm cleaning on my test environment, through the upstream posted state of the l2vni driver, using the local port attachments, its on distinct vlan segment attached to the vm's ports, booted, running, attached in ovn to a geneve network, to a router with packets flowing to and frow20:20
JayFTheJulia laid down a gauntlet, I think I'm going to have to pick it up20:21
JayFmaybe a good thing to discuss during ptg20:22
JayF(direct to placement)20:22
TheJuliaWe've discussed it before actually20:22
TheJuliajust, lacking push to close that loop since most people are not jumping from enrolled in ironic to deployable via nova20:22
TheJuliaand if you need to explicitly deploy metal, you can always just have ironic do it too20:23
TheJulianova just does the vifs because it is kind20:23
TheJuliato clarify, not jumping by meaning "not willing to wait the 5-ish minutes at worst"20:27
JayFMy main use case is twofold: 1) improving the failover story for nova-compute (right now an upgrade is guaranteed downtime of however long it takes a dormant nova-compute to get running again) and 2) reducing the lag time between a node exiting cleaning and being re-available to provision20:32
JayFI think directly updating placement would potentially have a positive impact in both cases, but I'm not 100% sure20:32
TheJuliaIt would allow the loop to be more of a lazy follow up20:42
TheJuliaWhich is a plus20:42
JayFprimarily I realized that since we changes the nove-compute HA model20:42
JayFit's impossible to do an HA-as-everyone-else-means-it openstack upgrade with ironic20:42
JayFbecause you're guaranteed downtime no matter how you roll (yes you can shard to compartmentalize, but the  value of that depends wildly on use case; it's not valuable to me)20:43
TheJuliaHmmm yeah, but truth be told, most folks take downtime on upgrades in some way/shape/form… even when they don’t need to because for the they are managing risk and set that expectation.20:44
TheJuliaKnowing/understanding/modeling differences evolve from there20:45
JayFI am not speaking so much in the theoretical as I am in the "no really, this annoys my downstream so I wanna fix it" :D 21:39
JayFbut we also tend to heavily use rebuilds-in-place which makes it so compartmentalized failure ha doesn't fit our use cases21:39
TheJuliaFair enough!22:29
opendevreviewClif Houck proposed openstack/ironic master: Filter out already attached portlikes in plan_vif_attach  https://review.opendev.org/c/openstack/ironic/+/97452122:46
opendevreviewClif Houck proposed openstack/ironic master: WIP: Update TBN simulator for vif_attach planning  https://review.opendev.org/c/openstack/ironic/+/97369122:46
opendevreviewClif Houck proposed openstack/ironic master: Filter out NoMatch actions in _vif_attach_tbn  https://review.opendev.org/c/openstack/ironic/+/97456922:46
clifcardoe: more corrections if you're around, but I get its late friday afternoon22:46
clifI think I've handled the problematic cases that have sprung to mind so far22:48
clifie, don't consider portlikes that are already attached, handle multiple actions per trait more correctly22:48
TheJulialast call for vxlan spec approvals22:50
TheJuliahttps://review.opendev.org/c/openstack/ironic-specs/+/95940122:50
JayFI'm OK to do a post-merge review, I don't have the context to get it now23:00
cardoeI'm good with it now. I think we'll flesh it out as we learn some other bits.23:05
opendevreviewDoug Goldstein proposed openstack/ironic master: make inspection doc link better in the side menu  https://review.opendev.org/c/openstack/ironic/+/97457023:07
cardoe^ that's another super boring one23:08
cardoehttps://review.opendev.org/c/x/proliantutils/+/974571 with that I'm gonna throw in the towel23:20
cardoeCause I hate every computer right now23:20
TheJuliaewww23:21
* TheJulia gives cardoe the certificate of computer hate and the martini of a quiet chill weekend23:21
cardoeI had asked him to submit it for a bit. That does fix ilo5 boxes.23:22
TheJuliathere are a few things wrong with that library, and if it wasn't for redfish I would have forked it long ago23:23
cardoeI don't doubt it.23:25
cardoeThat change allows me to smile and nod at the ilo5 hardware folks. I'd rather it be shared with the world.23:26
TheJulia++23:32

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