Tuesday, 2025-11-25

opendevreviewMerged openstack/ironic master: fix: remove unnecessary trailing slash  https://review.opendev.org/c/openstack/ironic/+/96815000:42
opendevreviewMerged openstack/ironic master: ci: remove snmp ci job  https://review.opendev.org/c/openstack/ironic/+/96321501:22
opendevreviewMerged openstack/ironic master: Configuration file for Trait Based Networking  https://review.opendev.org/c/openstack/ironic/+/96259801:44
opendevreviewMerged openstack/ironic master: chore: disable pycodestyle and pyflakes double runs  https://review.opendev.org/c/openstack/ironic/+/96815101:54
opendevreviewSteve Baker proposed openstack/ironic master: OciImageService detect bootc image  https://review.opendev.org/c/openstack/ironic/+/96676001:55
opendevreviewSteve Baker proposed openstack/ironic master: Swap to bootc deploy interface for bootc images  https://review.opendev.org/c/openstack/ironic/+/96676101:55
opendevreviewVerification of a change to openstack/ironic master failed: chore: fix spelling mistakes  https://review.opendev.org/c/openstack/ironic/+/96815203:41
opendevreviewMerged openstack/ironic master: chore: fix spelling mistakes  https://review.opendev.org/c/openstack/ironic/+/96815205:51
opendevreviewMerged openstack/ironic master: Remove inspector inspect interface  https://review.opendev.org/c/openstack/ironic/+/96747606:50
rpittaugood morning ironic! o/07:40
opendevreviewRiccardo Pittau proposed openstack/tenks master: Modernize and fix CI  https://review.opendev.org/c/openstack/tenks/+/96820008:53
opendevreviewRiccardo Pittau proposed openstack/ironic master: Update release mappings for new ironic release 33.0  https://review.opendev.org/c/openstack/ironic/+/96830609:53
rpittaupreparation for next release ^09:54
rpittauI put 33.0 to highlight the removal of inspector interface09:54
opendevreviewMerged openstack/ironic master: Fail flat driver rebind when no VIFs are bound  https://review.opendev.org/c/openstack/ironic/+/96777810:38
opendevreviewMerged openstack/ironic master: chore: bump to latest pre-commit tools  https://review.opendev.org/c/openstack/ironic/+/96815310:53
dtantsurrpittau: I assume you don't expect more API or RPC changes?11:58
dtantsurI know that alegacy's patches add a new entry to the mappings11:58
opendevreviewDmitry Tantsur proposed openstack/bifrost master: WIP add an OCI artifact registry  https://review.opendev.org/c/openstack/bifrost/+/96138812:38
alegacycid: TheJulia: how would you feel about if I do a:  s/portchannel/lag/  to rename that to a more vendor agnostic term.12:44
alegacysorry, didn't mean cid: there... i mean cardoe: ^^12:44
* cid nodes in the background12:45
cardoeWorks for me12:55
alegacyk, thanks12:55
rpittaudtantsur: I was planning to cut bugfix this week, will alegacy patches land in time?13:36
cardoerpittau: on stable branches?14:12
rpittaucardoe: on master14:13
cardoeah okay14:13
cardoeThe only show stopper on master for me is https://review.opendev.org/c/openstack/ironic/+/96825814:14
cardoeI have to update the tests because the tests validated the broken behavior.14:14
cardoehttps://review.opendev.org/c/openstack/ironic/+/968226 that'll fix the reported bug that cid triaged during the meeting yesterday14:16
TheJuliagood morning14:16
cardoewe landed the inspector removal so the bug fix release will be the first without inspector14:17
TheJuliaalegacy: I guess that works for me14:19
rpittauyeah I actually wanted to cut bugfix now because of that change :)14:19
TheJuliabecause of which change?14:19
rpittauTheJulia: the inspector interface removal14:19
TheJuliaseems like cardoe has good reason to hold 33.0's release until he has tests written to confirm/verify the broken heavior he has with images, if he can make it appear quickly :)14:20
cardoeLet me make coffee and I'll do it right now.14:20
TheJuliak14:21
cardoe100% the patch I did is correct and fixes the behavior. 100% the tests are wrong.14:21
cardoeWe attempt to read the source format type from the dest file which doesn't yet exist because we create the dest file a couple of lines down.14:21
cardoeThe problem is it's a spider web of "path" and "path_tmp" variables14:21
rpittauyeah, I... nvm :D14:21
rpittauI haven't proposed 33.0 yet, I was just updating the release mapping14:22
rpittauI don;t think cardoe's change influnces that14:22
cardoeThe tests just do an assert.called_with() and its testing that we read the source format from the dest file.14:23
TheJuliadtantsur: any issue if I rev your bifrost OCI registry change?14:25
opendevreviewJulia Kreger proposed openstack/bifrost master: WIP add an OCI artifact registry  https://review.opendev.org/c/openstack/bifrost/+/96138814:28
TheJulialets see!14:28
opendevreviewJulia Kreger proposed openstack/ironic stable/2025.2: Fail flat driver rebind when no VIFs are bound  https://review.opendev.org/c/openstack/ironic/+/96832614:39
opendevreviewJulia Kreger proposed openstack/ironic stable/2025.1: Fail flat driver rebind when no VIFs are bound  https://review.opendev.org/c/openstack/ironic/+/96833014:39
opendevreviewJulia Kreger proposed openstack/ironic bugfix/31.0: Fail flat driver rebind when no VIFs are bound  https://review.opendev.org/c/openstack/ironic/+/96833114:40
opendevreviewJulia Kreger proposed openstack/ironic bugfix/30.0: Fail flat driver rebind when no VIFs are bound  https://review.opendev.org/c/openstack/ironic/+/96833214:40
opendevreviewRiccardo Pittau proposed openstack/bifrost master: Remove tinyipa support and switch to DIB-based IPA  https://review.opendev.org/c/openstack/bifrost/+/96440414:47
opendevreviewRiccardo Pittau proposed openstack/bifrost master: Remove tinyipa support and switch to DIB-based IPA  https://review.opendev.org/c/openstack/bifrost/+/96440415:12
TheJuliadtantsur: fyi, next rev for your bifrost change already staged locally, which I suspect based upon the current failures will pass CI once proposed15:22
dtantsurTheJulia: sure, feel free to update the patch whichever way you want (just let me know when you're done)15:23
TheJuliadtantsur: Were you thinking a separate change to upload and use an artifact?15:23
dtantsurMy plans for next patches are: download ORAS client to /opt/bifrost/bin and finally switch the default deployment logic to use the registry instead of a file in httpd15:23
dtantsurheh, good timing :)15:24
dtantsuryes, I did not want to keep piling things into this already large change15:24
TheJuliafair enough, so if you want to start on a separate change then, you definitely can, looks like the current one is close15:24
dtantsurGot it. Time to exercise a bit, then I can task my Claude with the ORAS one15:26
TheJuliaheh, okay15:27
TheJuliaEnjoy15:27
* TheJulia makes a release note15:27
TheJuliaChanging your generated by to assisted by, btw.15:33
dtantsurfair, it has received enough human attention15:33
TheJuliaYeah, that is what I was thinking looking at it and the revs15:34
opendevreviewDoug Goldstein proposed openstack/ironic master: fix: use the correct path to the image when deep image inspection is off  https://review.opendev.org/c/openstack/ironic/+/96825815:50
cardoeThat should pass the tests. Sorry. I had a 1:115:51
cardoehttps://review.opendev.org/c/openstack/ironic/+/968226 if you guys wanna close out that bug15:52
opendevreviewJulia Kreger proposed openstack/bifrost master: add an OCI artifact registry  https://review.opendev.org/c/openstack/bifrost/+/96138816:09
opendevreviewJulia Kreger proposed openstack/sushy master: Don't require Boot and Actions for Systems  https://review.opendev.org/c/openstack/sushy/+/96771516:51
TheJuliahmmmmm17:01
opendevreviewDoug Goldstein proposed openstack/ironic master: Set upper_bound column in bios_settings table to bigint  https://review.opendev.org/c/openstack/ironic/+/96834817:01
TheJuliadtantsur: so slight problem, keystone and the registry both want port 500017:03
dtantsurlol17:08
TheJuliayeaaah... 17:08
dtantsurTheJulia: it should not be a big deal to redirect the registry to a different port17:08
TheJuliaokay, I'll change it to $something17:08
dtantsurTheJulia: can we have a port in oci:// URLs?17:08
TheJuliaafaik, yes17:08
TheJuliakeep in mind, quay runs as just a normal https service17:09
TheJuliaso you would have had to explicitly define 5000 to use it as is today17:09
dtantsuroh yeah, true, so even 5000 would need to be included explicitly17:09
dtantsurright17:09
dtantsuranyway, if it does not work, we fix it :)17:09
TheJuliaok17:11
opendevreviewMerged openstack/ironic master: Update release mappings for new ironic release 33.0  https://review.opendev.org/c/openstack/ironic/+/96830617:11
opendevreviewJulia Kreger proposed openstack/bifrost master: add an OCI artifact registry  https://review.opendev.org/c/openstack/bifrost/+/96138817:14
cardoeDoes conductor need to host the files locally in its http server with OCI?17:16
dtantsurcardoe: not necessarily, depending on image_download_source17:16
dtantsurTheJulia: re docs: I was thinking about an update to https://docs.openstack.org/bifrost/latest/user/architecture.html (I can do it in a follow-up if you run out of cycles)17:16
TheJuliadtantsur: The more I was thinking about it, the more I was thinking it might just be best as a follow-up or with the actual OCI usage chagnes17:17
TheJuliajust so we paint the entire picture at once17:18
cardoeSo we generate a url which doesn’t need auth?17:18
dtantsurI think the conductor passes auth to IPA together with the URL. I don't remember for sure, TheJulia?17:18
TheJuliaIf auth is required, we're able to pass the session token out to the agent. Althoguh, standalone auth wise using with an htpasswod, I have no idea because its a digest auth, nor an actual session auth with a shared token17:19
cardoeI would love to get rid of conductor’s http server.17:22
TheJuliaThe conductor's local http server is still required for conductor side local caching and ipxe17:23
dtantsuryeah, we're not quite there yet. Virtual media also serves the ISO locally.17:23
* dtantsur claude goes brrrrr17:24
TheJuliaYeah, there is basically no getting rid of it short of kicking entire use cases to the curb17:24
dtantsurThe biggest conundrum IMO is that both BMCs (for virtual media) and iPXE can be very picky about the source of their downloads.17:25
dtantsurOtherwise, we'd be able to just serve a virtual media ISO directly to the BMC17:25
TheJuliaIts a challenge of the range of use cases, and what makes ironic powerful is a variety, not any singular pattern one must fit into17:31
TheJuliaGranted, specific flows/patterns/tool/approaches may do some of the needful and eliminate the need in some cases, but that will always be a specific pattern and configuration which may not apply at all to what someone else needs to do17:32
TheJuliaso technically, one could upload the vmedia iso to a swift registry as well, but we also know some vendors don't like things like & and ? in those URLs either. :\17:34
dtantsurThe ORAS patch is ready locally, are you by chance updating the parent patch now?17:34
TheJuliawhat is in CI should be good since I changed the port to 6000 on it17:34
TheJuliaits running presently17:35
dtantsurah, I can probably use --no-rebase to avoid changing it.. or not?17:35
TheJuliayou can17:35
opendevreviewDmitry Tantsur proposed openstack/bifrost master: bifrost-registry-install: install the ORAS client  https://review.opendev.org/c/openstack/bifrost/+/96835517:35
dtantsurin any case, it worked. let's see what the CI says (I still don't have a working local bifrost, sigh)17:36
TheJuliaLGTM, next would be upload I guess17:36
dtantsuryeah, it will require a bit more thinking than just telling claude what to do17:37
TheJuliayeah17:37
opendevreviewClif Houck proposed openstack/ironic master: WIP: Add configuration options for trait based networking  https://review.opendev.org/c/openstack/ironic/+/96805417:37
dtantsurTheJulia: TLS settings could be a potential headache17:39
TheJuliaThat is definitely a thing17:40
dtantsuroh, and I wonder if port 6000 is a good choice: https://www.speedguide.net/port.php?port=600017:40
TheJulia5500 ?17:43
dtantsurTheJulia: we could even do 5050 to confuse people migrating from inspector :-P17:43
dtantsur(seriously though, maybe reusing a previously known code is not a terrible idea: it may already be enabled on firewalls)17:46
dtantsurs/code/port/ but you got it17:47
TheJuliaThat is just evil17:48
TheJuliaThat being said its looking good so far for 6000. I also fixed some of the value usage to use templated values17:49
dtantsurhttps://docs.openstack.org/ironic/latest/admin/oci-container-registry.html doesn't say much about the TLS situation, we need to add it17:54
TheJuliaAIUI, non-https usage is actually not supposed to work for that stuff17:56
TheJuliaits *all* expected to be httpsified at all times17:56
dtantsuryep, but the only configuration we have is webserver_verify_ca17:56
dtantsurand even there we only accept a boolean and crash if a path is provided :)17:57
dtantsurI want to do 2 things: first have an option specific to oci (with a fallback to webserver_verify_ca), second also support paths17:58
* dtantsur hack hack hack17:58
TheJuliareasonable I guess for self signed18:01
opendevreviewDmitry Tantsur proposed openstack/ironic master: Improve TLS support for OCI images  https://review.opendev.org/c/openstack/ironic/+/96836118:09
dtantsursomething like this ^^18:10
opendevreviewMerged openstack/ironic stable/2025.2: fix: glance image member lookup resulted in an empty list always  https://review.opendev.org/c/openstack/ironic/+/96822618:11
dtantsurI wonder if we need per-node options too. I think we do, especially for container cases where modifying the global configuration is not trivial.18:12
opendevreviewDmitry Tantsur proposed openstack/bifrost master: Enable TLS by default, deprecate enable_tls  https://review.opendev.org/c/openstack/bifrost/+/96836318:29
dtantsurNot sure what made me do ^^. Probably perfectionism.18:29
TheJuliaseems reasonabkle, to me18:30
TheJuliareasonable18:30
cardoehttps://review.opendev.org/c/openstack/ironic/+/968258 there's the passing test TheJulia 18:37
TheJuliabrraaaains18:39
TheJuliasorry, my uncle called me and gave me a flurry of details18:39
cardoeJust pinging cause that was the patch we wanted to land before rpittau made the bugfix release.18:40
dtantsurI wonder if there is a mirror of the registry container, so that we don't hit docker's limits https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_5cb/openstack/5cb8fe7cd9974fcfae0c2edd9eed5e2f/logs/bifrost-registry.log18:50
TheJuliauuugh18:53
dtantsurI wonder if quay is easy to install18:56
TheJulia... I was thinking a similar question18:56
* dtantsur checking https://docs.projectquay.io/welcome.html18:56
dtantsurugh, it has a lot of requirements18:57
TheJuliabut there are containers...18:58
TheJuliaexcept the flow is you have to go create an initial account on the webui18:59
dtantsursigh18:59
dtantsurokay, I wonder if we can just install the registry binary from https://github.com/distribution/distribution/releases (similarly to my ORAS patch)18:59
TheJuliaThat I'm skeptical of19:02
dtantsurTheJulia: okay, so we can use the binary from ^^ which has the benefit of not messing with bloody podman. Or we use the mirror in the CI https://quay.io/repository/opendevmirror/registry?tab=tags19:02
TheJuliagiven the explicit redis requirement19:02
dtantsursorry, I'm confusing things. the former docker's registry (now distribution) is a single binary19:02
TheJuliaoh, wait, there is a mirror already19:02
dtantsurand yes, there is a mirror apparently19:02
TheJuliaoh, okay19:02
* TheJulia holds up the use the mirror sign19:02
TheJuliaSorry, my uncle called me a little while ago and decided to provide. "here is everything going on" discussion, so I'm mentally derailed at the moment19:03
dtantsurIt's too late for my brain, I see benefits in both options, so I'll leave it up to you I guess19:03
dtantsurah, so you too. we can get back to it tomorrow19:03
TheJuliaGet some rest, I'm going to go get some lunch and stop by the grocery store, and then get back to it19:03
dtantsurenjoy your lunch!19:04
opendevreviewDoug Goldstein proposed openstack/networking-generic-switch master: WIP: l2vni plug case  https://review.opendev.org/c/openstack/networking-generic-switch/+/96837719:24
cardoeTheJulia: ^ yolo19:24
cardoeI did cisco since that's what ya got but I also think the command might need a name19:26
opendevreviewMerged openstack/bifrost master: Add support for OVS as a virtual switch  https://review.opendev.org/c/openstack/bifrost/+/96203820:21
TheJuliacardoe: but what happens when that port goes away? :)21:38
cardoeyou throw away all the hardware and start over?21:41
cardoeI'll wire it up but I was just trying to say it's not a big change.21:41
cardoeOnce again... https://bugs.launchpad.net/networking-generic-switch/+bug/2114451 makes it hard21:41
cardoeCause you have to walk all the ports to see what segments are in use21:42
cardoeCause you need to know if that vlan is in use on that switch21:47
cardoeSo you need to know if there's any ports on that segment.21:47
TheJuliaOh, I know its not a big change21:48
TheJuliaI know its relatively small, the key is we need to also do the deletes/cleanup as needed which somehow needs to be aware in that model. Maybe we can figure out a way or not21:48
TheJuliaand yeah, that is an awful bug21:48
cardoeThere actually might be an escape hatch here.22:00
TheJuliaYeah, I concur, just can't look for it right now22:01
opendevreviewJulia Kreger proposed openstack/bifrost master: add an OCI artifact registry  https://review.opendev.org/c/openstack/bifrost/+/96138823:00
opendevreviewJulia Kreger proposed openstack/bifrost master: bifrost-registry-install: install the ORAS client  https://review.opendev.org/c/openstack/bifrost/+/96835523:00
opendevreviewJulia Kreger proposed openstack/bifrost master: WIP: Upload a disk image to OCI  https://review.opendev.org/c/openstack/bifrost/+/96841623:00
opendevreviewJulia Kreger proposed openstack/bifrost master: WIP: upload and use the OCI image  https://review.opendev.org/c/openstack/bifrost/+/96841723:00
TheJuliadtantsur: claude attempts, change at will23:04
opendevreviewSteve Baker proposed openstack/ironic master: Swap to bootc deploy interface for bootc images  https://review.opendev.org/c/openstack/ironic/+/96676123:06
opendevreviewDoug Goldstein proposed openstack/ironic master: fix: ensure that the bootloader path exists  https://review.opendev.org/c/openstack/ironic/+/96841823:12
opendevreviewDoug Goldstein proposed openstack/ironic master: fix: do not allow nested paths in loader_file_paths  https://review.opendev.org/c/openstack/ironic/+/96841923:12
cardoeThe first one is a backport and the 2nd updates to the intended behavior23:14
TheJuliaThanks cardoe !23:33
opendevreviewKaifeng Wang proposed openstack/ironic-specs master: Flow control on PXE related port requests  https://review.opendev.org/c/openstack/ironic-specs/+/94796123:53
opendevreviewJulia Kreger proposed openstack/networking-generic-switch master: DNM: Enable NGS to auto-prune unused vlans  https://review.opendev.org/c/openstack/networking-generic-switch/+/96842023:55
TheJuliacardoe: super-duper rough idea, but something ^^^^ could be a basis to unwind a vlan out of existence on detachments23:56

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