Wednesday, 2025-09-03

opendevreviewKaifeng Wang proposed openstack/ironic-python-agent-builder master: Exclude unnecessary firmewares  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/95925500:37
opendevreviewMaksim Malchuk proposed openstack/bifrost stable/2025.1: Do not pass empty values to instance_info  https://review.opendev.org/c/openstack/bifrost/+/95927605:31
opendevreviewMaksim Malchuk proposed openstack/bifrost stable/2024.2: Do not pass empty values to instance_info  https://review.opendev.org/c/openstack/bifrost/+/95927705:32
opendevreviewMaksim Malchuk proposed openstack/bifrost stable/2024.1: Do not pass empty values to instance_info  https://review.opendev.org/c/openstack/bifrost/+/95927805:32
abongalegood morning ironic o/08:14
opendevreviewTakashi Kajinami proposed openstack/ironic-inspector master: sqlalchemy: Use built-in declarative  https://review.opendev.org/c/openstack/ironic-inspector/+/95376409:04
TheJuliagood morning12:35
opendevreviewMaksim Malchuk proposed openstack/bifrost stable/2024.1: [2024.1 only] Drop upgrade jobs since 2023.2 brach doesn't exists  https://review.opendev.org/c/openstack/bifrost/+/95931712:39
opendevreviewMaksim Malchuk proposed openstack/bifrost stable/2024.1: Do not pass empty values to instance_info  https://review.opendev.org/c/openstack/bifrost/+/95927812:40
clifo/13:30
opendevreviewPierre Riteau proposed openstack/bifrost master: [WIP] Test CentOS 10 support  https://review.opendev.org/c/openstack/bifrost/+/95885314:03
opendevreviewGrzegorz Koper proposed openstack/networking-generic-switch master: Support 802.3ad port groups on Arista devices  https://review.opendev.org/c/openstack/networking-generic-switch/+/95911814:31
dtantsurTheJulia: hi! what was driving the limitation on deploy_kernel/deploy_ramdisk in https://docs.openstack.org/ironic/latest/admin/oci-container-registry.html#available-url-formats?14:32
dtantsurIt was just raised as a potential adoption blocker in the Metal3 meeting14:32
TheJuliaI don't remember offhand14:33
TheJuliadid I note it in the code?14:33
TheJulia(On 2nd of ?4? hour long meetings this morning)14:33
dtantsurNot sure, I've just learnt about this limitation14:36
JayFyou'd have to update the tag atomically across two things  (x_kernel/x_ramdisk) to avoid issues14:38
JayFI could see it being logical to block off that route14:38
dtantsurI'm personally not terribly worried, compared to inability to use tags14:39
JayFhttps://review.opendev.org/c/openstack/ironic/+/937896 I haven't found the limitation code14:42
JayFbut that's the commit14:42
opendevreviewClif Houck proposed openstack/ironic master: Add a new 'physical_network' field to the Portgroup object  https://review.opendev.org/c/openstack/ironic/+/95562515:00
TheJuliaWOOHOO... an hour break15:01
TheJuliabut I'm still on a meeting15:01
TheJuliadtantsur: I think someone just needs to try it out at this point. I think the base issue is more to what JayF was noting. It might be something early on that got removed through iteration15:11
TheJuliadtantsur: I *think* it might be tags require authentication, but I think it might just work based upon the current code state15:27
TheJuliaOh, looks like I need to revise those docs anyhow15:31
dtantsur++ to just trying, yeah15:31
TheJuliaYay for super complicated features!15:31
dtantsur\o/15:31
TheJuliaset it up with quay, easiest path15:31
dtantsurIf it makes you feel better, it opens a lot of potential for metal315:31
dtantsurLike, getting rid of this whole ipa-downloader stuff (if you know)15:31
dtantsurand from ad hoc image caches15:31
TheJuliathe OCP image service doesn't support tags either, fwiw.15:32
dtantsurYeah, you mentioned. But this all is in the upstream context, so I don't care.15:32
TheJuliayeah15:32
dtantsur(we also research OCI images for a feature downstream, but there we don't care about kernel/ramdisk)15:32
TheJuliaI only very early did a bunch with quay, and then shifted to the more restricted OCI image service15:33
dtantsurOCP you mean? I found it very annoying, this image service.15:34
TheJuliayeah, OCP15:34
TheJuliasorry, BRRRAAAINS15:34
dtantsurEverything around authentication is just.. bizzare15:34
* dtantsur -> quick walk15:34
TheJuliaIt... makes sense. kind of.15:35
TheJuliaWe're getting pressure to ensure our releases for End of Cycle, we should dot our I's and cross our T's16:41
TheJuliacardoe: I met with some of our neutron folks today. I think we're going to need a spec, at least maybe in just the ironic context, which would talk about the rest of the picture in a standardized model. I do think they finally get the high level flow understood, but there are some questions and beginning to pin down another part of the known allows us to focus on the unknowns as well.17:08
TheJuliacardoe: I *suspect* I may be doing that unless I can somehow convince someone else to draft something17:08
cardoesounds good. I really then need to sit down and hack on the spec some more.17:09
cardoeoh speaking of dotting Ts and crossing Is...17:10
TheJuliaFor those watching, this is in regards to attaching baremetal nodes to vxlan networks17:10
JayFI assume at some point we'll have to tackle the bigger problem, e.g. "model baremetal network concepts in neutron" starting with portgroups/bonding17:10
cardoeI've got a further revision to also say "vxlan attach or vlan attach" since the typical case is gonna be vlan attach.17:10
JayFI don't know how associated those issues are17:11
TheJuliaI think portgroups are just super weird and maybe more noise for them, I dunno17:11
TheJuliaits worth a discussion because they are aware it exists17:11
cardoeJayF: so it works and its werid17:11
cardoeBasically you still have one neutron port but it's binding info is a list instead17:11
cardoeCause it's all just JSON free form being passed around willy nilly17:12
TheJuliaI guess the open question for something else I'm needing to look at is... how do I know what the portchannel name is in the switch to appropriately tag vlans to?17:12
TheJuliaSpecifically because Cisco gear you have to do vlans on the port channel interface, not the port.17:12
TheJuliaWhere as some other vendors, its the inverse17:12
cardoeSo technically the list is in a JSON object. We're slapping the portgroup info into that.17:13
TheJuliaYeah17:13
cardoeAnd something I pitched at neutron and got told to write a spec.17:14
TheJuliaThere is definitely some bridges to build there, in the form of shared context... and actual bridges17:15
* TheJulia ducks17:15
cardoeSo one other thing I had on the redfish stuff... rather than all the hardware gathering in one big inspect function. I broke out each piece into _get_$THING_info(). And then drivers like idrac-redfish can override that if they want.17:15
cardoeBut I don't love that so the other horrible idea I've had is to share the system sushy object to all the hooks in some fashion and allow hooks to be written to gather and populate more fields.17:16
TheJuliaI think there a mixed bag of challenges since some of the BMCs take like a minimum of 20 seconds to return something usable just for one of those areas.17:18
* TheJulia makes lunch before trying to do anything else17:27
TheJuliaIs there anything else we feel we must get merged before we release $allthethings ?17:27
TheJulia^^^ ironic-cores17:27
TheJuliasuper easy logging improvement: https://review.opendev.org/c/openstack/ironic/+/95582117:47
opendevreviewJulia Kreger proposed openstack/ironic master: Revert "Make ironic-tempest-uefi-redfish-vmedia-4k non-voting"  https://review.opendev.org/c/openstack/ironic/+/95858017:50
TheJuliaLooking at the status dashboard, I think we need to go ahead and just cut the release17:52
*** Guest22851 is now known as diablo_rojo_phone18:12
TheJuliaAnyone done portgroups with ironic and have the context relatively fresh in their minds?18:22
TheJulianvmd, found the code18:26
opendevreviewVerification of a change to openstack/ironic-python-agent master failed: Support transport type as a root device hint  https://review.opendev.org/c/openstack/ironic-python-agent/+/95574219:21
opendevreviewVerification of a change to openstack/ironic-python-agent master failed: Support transport type as a root device hint  https://review.opendev.org/c/openstack/ironic-python-agent/+/95574219:52
opendevreviewJulia Kreger proposed openstack/ironic master: Add a prelude for 2025.2  https://review.opendev.org/c/openstack/ironic/+/95938920:51
opendevreviewMerged openstack/ironic master: Revert "Make ironic-tempest-uefi-redfish-vmedia-4k non-voting"  https://review.opendev.org/c/openstack/ironic/+/95858021:18
opendevreviewMerged openstack/ironic-python-agent master: Support transport type as a root device hint  https://review.opendev.org/c/openstack/ironic-python-agent/+/95574222:24
TheJuliawoot22:51
opendevreviewJulia Kreger proposed openstack/ironic master: Add a prelude for 2025.2  https://review.opendev.org/c/openstack/ironic/+/95938923:43
opendevreviewJulia Kreger proposed openstack/ironic-specs master: WIP VXLAN networking  https://review.opendev.org/c/openstack/ironic-specs/+/95940123:45
TheJuliacardoe: ^^^ to try and begin to solidify the discussion a bit more23:46
cardoeThank you. I literally was just thinking about this.23:47
cardoeFinding shelter from the rain in Nashville.23:48
cardoe+2’d https://review.opendev.org/c/openstack/ironic/+/955821 for the release.23:50

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