Thursday, 2026-06-25

opendevreviewSteve Baker proposed openstack/ironic-python-agent-builder master: Fix codespell job  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/99483101:19
opendevreviewMerged openstack/ironic-python-agent-builder master: Fix codespell job  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/99483103:48
*** hroy_ is now known as hroy03:59
rpittaucardoe: would really love an answer for https://review.opendev.org/c/openstack/networking-generic-switch/+/994433 when you got a moment, as I wrote I'm ok removing just neutron (and it probably makes more sense), thanks! :)07:57
opendevreviewRiccardo Pittau proposed openstack/ironic-prometheus-exporter stable/2025.1: [stable] only] Replace pkg_resources with pathlib for resource file loading  https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/99476808:34
opendevreviewDmitry Tantsur proposed openstack/ironic master: Modernize the Metal3 functional jobs  https://review.opendev.org/c/openstack/ironic/+/99260711:38
dtantsurJayF: hey, I recall you didn't feel well about a service step to attach an arbitrary ISO. Do you have the same feelings about a clean step?12:53
cardoerpittau: oh. I didn’t -1 so it’s good merge it.12:58
cardoeIn the future I wanna nudge them to have smaller deps.12:58
dtantsurcardoe: oh hi, while we have you here: you mentioned you had some thoughts on hardware discovery, didn't you?13:00
dtantsurdo you have similar requests or something brewing already?13:01
rpittaucardoe: ok, thanks, but thinking about what you wrote it's probably better to remove just neutron for the time being :)13:07
opendevreviewRiccardo Pittau proposed openstack/networking-generic-switch master: Make neutron an optional dependency  https://review.opendev.org/c/openstack/networking-generic-switch/+/99443313:13
cardoerpittau: agreed13:28
cardoedtantsur: got something for me to give feedback on?13:28
cardoeI’m in the woods currently so not caught up on context.13:30
dtantsurNo, not yet, but may in a couple of hours13:30
dtantsurit's re your comment on the midcycle etherpad13:30
rpittauCI is finally happy here \o/ https://review.opendev.org/c/openstack/bifrost/+/99112813:46
JayFdtantsur: My objection was primarily related to booting a completely unmanaged by ironic workload on an active machine. So I think it should be okay as a clean step.13:57
dtantsurGot it. I'll see what these people actually want.13:58
cardoedtantsur: ah yes. can we chat next week?14:22
dtantsurcardoe: sure! More reasonable weather can help me think :)14:23
dtantsurJayF: so, apparently some vendors refuse to implement firmware updates via Redfish, providing an ISO instead :(14:32
JayFI mean, sounds like a vendor who shouldn't have their hardware purchased.14:39
dtantsurOh yeah, it's on my personal naughty list already. But I'm not the one making purchases..14:40
JayFOr that the cloud users need to do some engineering work to extract the actual firmware writes from the ISO and put it into IPA in-band14:40
JayFI don't think we should compromise our security stance because a vendor is being obtuse. 14:40
JayFI am getting more and more convinced that being willing to compromise for vendors contributes to the problem14:40
dtantsurWe're leaning towards the same downstream, but it's never an easy conversation14:41
JayFdtantsur: There is some in-band standardization for firmware updating in linux -- mainly fwupd14:56
JayFdtantsur: the big downside there is that many vendors' firmware bundles used for fwupd (e.g. almost all the framework ones) are EFI-based14:56
JayFso it's just doing efi boot stuff to make the firmware update happen14:56
TheJuliaits actually the best case in-band model since there are guard rails built in14:59
TheJuliaFWIW14:59
JayFyou mean fwupd or EFI based or both?14:59
TheJuliaEFI based model15:03
TheJuliathat is functionally what the BMC model is doing in some cases15:03
TheJuliapresenting a virtual thumb drive with the file15:03
TheJuliaAt least for parts and pieces which can't be bit-banged or directly accessed15:03
TheJuliabit-banged is being avoided in favor of proper A/B or slotting models15:04
TheJulialike... one vendor wants to hold up to 8 firmware images for some devices15:04
TheJuliaActive, Secondary, Tertiary 1-615:04
TheJuliaJayF: That is a most excellent email.15:05
opendevreviewMerged openstack/ironic master: docs: Augment bootc docs  https://review.opendev.org/c/openstack/ironic/+/99342115:50
opendevreviewDmitry Tantsur proposed openstack/ironic master: Modernize the Metal3 functional jobs  https://review.opendev.org/c/openstack/ironic/+/99260716:18
JayFhttps://review.opendev.org/c/openstack/ironic-specs/+/992741/2#message-3def7fa9fcbaa5d33b59c26ebe2a87e62f0ab4a2 please take a look at the container interface proposed here17:03
JayFit's pretty simple spec and almost got a +2 from me on first pass17:03
opendevreviewMerged openstack/ironic master: docs: Add a Getting Started map to the landing page  https://review.opendev.org/c/openstack/ironic/+/99477017:07
TheJuliaJayF: you -1'ed it... unless its been revised somewhere.17:35
TheJuliaand my browser is still showing old records17:35
JayFalmost :)17:36
JayFthe -1 is for naming17:37
JayFbut the implementation itself is solid17:37
JayFThe names of the interfaces there describe the method for spinning up the container, not the container engine17:37
TheJulia... true17:41
* TheJulia glares at claude for saying "I ran pep8" when it failed doing so17:44
JayFmy favorite is when it goes to make a venv to run pep817:48
JayFinstead of using tox like it says to do in the claude.md, personal instructions, and examples earlier in the chat17:48
JayFclaude models get dumber over time, I'm completely convinced of it17:48
JayFand the version numbers are fantasies17:49
TheJuliaYeah, and also the "oh, this looks like super cheap model can answer it" logic overriding user selections17:52
JayFI worked this weekend on improving my local email (I now sync to a local IMAP proxy + use imapfilter on office365 for filtering, it's great)17:53
JayFand mid-project, claude compacted the giant conversation we had about it17:53
JayFand it was like self-lobotomization. Suggesting things we knew wouldn't work. Making tradeoffs on my requirements without asking.17:53
JayFI was more pissed off than I've been at an AI in a while, just because it was so useful and became worse-than-useless in the blink of an eye, completely outside of my control17:53
TheJuliaYeah, I've started to keep an eye on the remaining context at times and explicitly clear it.17:54
JayFI almost wish I could "sticky" context?17:54
JayFlike in a spreadsheet, you save the top 10 rows or something so you only scroll the rest17:55
TheJuliahttps://github.com/anthropics/claude-code/issues/49861 :(17:55
JayFWhen I think about it, I tend to have it "self-compact" at that point by telling it to write me a prompt to start on (implementation, step 2, etc) so I can edit the prompt and make sure it didn't drop naything useful17:55
TheJuliaYeah, I've tried to drive it twoards very stepwise ordering for my sanity17:56
JayFWriting an issue in the claude-code github is about as useful as writing your issue on a post-it note, giving it a little kiss, shredding it, and spreading the shreds on the wind17:56
TheJuliaI did that with deferred task stuff pretty heavily and its not awful now thatI'm going through and reviewing the wip changes in fine detail and cleaning them up.17:56
* TheJulia gets out the giant post-it notes17:57
* TheJulia also gets out the torch17:57
opendevreviewJulia Kreger proposed openstack/ironic master: Add deferred actions data model, DB API, and versioned object  https://review.opendev.org/c/openstack/ironic/+/98922418:21
opendevreviewJulia Kreger proposed openstack/ironic master: Add config options, reschedule exception, and action creation helper  https://review.opendev.org/c/openstack/ironic/+/98922518:21
opendevreviewJulia Kreger proposed openstack/ironic master: Add two-step deferred actions cleanup DB methods  https://review.opendev.org/c/openstack/ironic/+/99494818:21
opendevreviewMerged openstack/ironic stable/2025.1: Add operator-configurable step disallow lists  https://review.opendev.org/c/openstack/ironic/+/99273218:38
opendevreviewJulia Kreger proposed openstack/ironic master: Add deferred action launcher periodic and executor  https://review.opendev.org/c/openstack/ironic/+/98922619:13
opendevreviewJulia Kreger proposed openstack/ironic master: Add deferred action cleanup periodic and stuck recovery  https://review.opendev.org/c/openstack/ironic/+/98922720:13
TheJuliacid: if you want to carry on rebasing the changes on that series while I'm on vacation, your welcome to.20:13
cidI will poke at it if I get a window.20:21
JayFat this rate he might be writing enhanced maintenance20:21
cid Enjoy the break!20:21
JayFthat scheduled maintenance window thing20:21
JayFso obvious but how did I never think of it20:21
TheJuliaoh yeah, totally makes so much sesne20:39
TheJuliabecause I've literally had to shutdown transformers20:39
TheJulia"oh noes! that takes row 3/4 out"20:39
TheJuliaI'm basically engaging the cloaking device and heading towards vacation20:44
JayFI will see you (literally!) Tuesday.20:47
TheJuliaWoot!21:20
opendevreviewSteve Baker proposed openstack/ironic-python-agent master: Enable centos10 publish job  https://review.opendev.org/c/openstack/ironic-python-agent/+/99257121:26
opendevreviewSteve Baker proposed openstack/ironic master: Switch jobs to build centos10 IPA images  https://review.opendev.org/c/openstack/ironic/+/99257321:49
opendevreviewSteve Baker proposed openstack/ironic-python-agent master: Switch ironic-standalone-ipa-src to centos10 ipa  https://review.opendev.org/c/openstack/ironic-python-agent/+/99498221:52

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