Monday, 2026-01-19

rpittaugood morning ironic! o/07:21
abongalegood morning o/09:03
opendevreviewcid proposed openstack/ironic master: Support `disable_ramdisk` during servicing  https://review.opendev.org/c/openstack/ironic/+/97301610:06
opendevreviewcid proposed openstack/ironic-python-agent master: Fix lookup proceeding with incomplete data  https://review.opendev.org/c/openstack/ironic-python-agent/+/97377610:06
dtantsurcardoe: hey, what about we merge https://review.opendev.org/c/openstack/ironic/+/966344 and ask janders to follow-up? The issue with largely broken upgrades is quickly heating up for us..11:08
opendevreviewElod Illes proposed openstack/ironic master: Adapt bindep for debian-trixie  https://review.opendev.org/c/openstack/ironic/+/97381812:44
opendevreviewMerged openstack/bifrost master: Provide a means to disable the registry service install  https://review.opendev.org/c/openstack/bifrost/+/97367312:54
janderscardoe sorry for not getting back to you earlier, been down with COVID last week13:08
janderswill get onto the follow-up this week for sure13:08
cardoedtantsur: yeah that works13:31
cardoejanders: sorry to hear but glad you are doing better13:32
janderscardoe thank you! yeah back to the living this week13:36
janderslast week was a bit of a shutdown -h now13:36
opendevreviewVerification of a change to openstack/ironic master failed: Make post-firmware-update reboot conditional on component  https://review.opendev.org/c/openstack/ironic/+/96634413:57
opendevreviewVerification of a change to openstack/ironic master failed: Fix firmware update failure when task remains in RUNNING state  https://review.opendev.org/c/openstack/ironic/+/97099813:57
opendevreviewVerification of a change to openstack/ironic master failed: Fix BIOS firmware not flashing when task completes before first poll  https://review.opendev.org/c/openstack/ironic/+/97099913:57
opendevreviewVerification of a change to openstack/ironic master failed: Fix BMC resource validation to catch all sushy exceptions  https://review.opendev.org/c/openstack/ironic/+/97100013:57
TheJuliagood morning14:01
dtantsurmorning TheJulia14:04
dtantsurhuh, what does CI hate today?14:04
TheJuliadon't know, is it us today?14:10
dtantsurLooks like it's DNS (tm)14:13
opendevreviewVerification of a change to openstack/ironic master failed: Make post-firmware-update reboot conditional on component  https://review.opendev.org/c/openstack/ironic/+/96634414:34
opendevreviewJacob Anders proposed openstack/python-ironicclient master: Add support for node health status field  https://review.opendev.org/c/openstack/python-ironicclient/+/96705514:53
TheJuliadtantsur: fwiw, I +2'ed the py313 patch, We can always follow it up14:55
TheJuliacardoe: will you have at least a little time to discuss vxlan today?14:56
kubajjo/15:00
janderso/15:01
TheJuliameeting time?15:02
TheJulia#startmeeting ironic15:03
opendevmeetMeeting started Mon Jan 19 15:03:00 2026 UTC and is due to finish in 60 minutes.  The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot.15:03
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:03
opendevmeetThe meeting name has been set to 'ironic'15:03
rpittauo/15:03
* TheJulia hands everyone coffeeeeee15:03
dtantsurI think we need to get back to asking the question "who is chairing the meeting next time?" :)15:03
dtantsuro/15:03
kubajjo/15:03
TheJuliadtantsur: ++ c oncur15:03
TheJuliaGreetings everyone, welcome to our weekly Ironic meeting. Our agenda can bef ound at https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_January_19.2C_202515:03
cido/15:03
TheJuliaerr, be found15:05
TheJuliaClearly, I need more coffee15:05
* TheJulia sips coffee and proceeed15:05
TheJulias15:05
TheJulia#topic Announcements / Reminders15:06
TheJuliaAs a standing reminder to review #ironic-week-prio items.15:06
TheJulia#link https://tinyurl.com/ironic-weekly-prio-dash15:06
TheJuliaWe have 30 items, lets get it below 25 this week.15:06
TheJulia(preferably 20)15:06
cid++15:07
rpittauif CI collaborates I think it's doable :)15:07
TheJuliaThis is week R-10 on the release cycle.15:07
TheJulia#link https://releases.openstack.org/gazpacho/schedule.html15:07
TheJuliaAs a reminder, we should check the cycle priorities as well. If you get a chance this week, please take a quick glance at that list.15:07
TheJuliaDoes anyone have anything else to remind us of or announce this week15:08
rpittauyup, I'm planning a round of releases btw15:08
TheJuliaDoes anyone have anything else?15:08
TheJuliaOnward15:09
TheJulia#topic Working Group Updates15:09
cardoeTheJulia: via msg? I’m driving home today15:10
TheJuliaFirst up standalone networking. alegacy you around today?15:10
TheJuliacardoe: could work15:10
TheJuliaMy impression is we're largely just in review phase for the standalone networking work.15:11
TheJuliadtantsur: An update for Async IO ?15:11
dtantsurNot much new on the async front. Still finishing the spec.15:12
dtantsur(re standalone networking, I've left some comments on the first patch)15:12
TheJuliaokay then15:13
TheJuliaAs for VXLAN networking, I think we're starting to make progress. I've gotten bit of a development test lab up and we have patches posted for networking-generic-switch to handle quite a bit of switch side configuration as it relates to the topic.15:14
TheJuliaThe spec still needs reviews, but I think we're in a state of consensus at this point15:14
TheJuliaDoes anyone else have anything working group related?15:14
cidNot at the moment15:14
TheJuliaSince we have no defined discussion topics, lets go onward to the Bug Deputy Updates15:15
TheJuliacid, the microphone is yours15:16
cidSure15:16
cidThere were a total of 6 new bugs, and 2 RFEs. 15:16
cid- Allow abort verb on DEPLOYWAIT15:16
cid- Deferred Tasks - Have a way to centralize periodics and minimize sleep usage15:16
cidI don't think these need reviews (?)15:16
TheJuliaThe deploywait one, I think we already agreed to15:16
TheJuliaDeferred tasks is for the proposed spec document which I think I updated last week15:16
cidIs it ready for reviews yet?15:17
TheJuliaYes, it is up for review in ironic-specs15:17
cid+++, cool cool15:17
TheJuliais that 6 new bugs across all ironic projects?15:17
cidIndeed, Bifrost, Ironic, Sushy, ...15:18
TheJuliaokay15:18
cid12th - 19th January, 202615:18
cid- https://bugs.launchpad.net/ironic/+bug/2138310 (in Progress, no review necesssary)15:18
cid- https://bugs.launchpad.net/bifrost/+bug/213737615:18
cid- https://bugs.launchpad.net/sushy-tools/+bug/213738715:18
cid- https://bugs.launchpad.net/ironic/+bug/213757115:18
cid- https://bugs.launchpad.net/ironic/+bug/213812915:18
cid- https://bugs.launchpad.net/python-ironicclient/+bug/213846615:18
cid- https://bugs.launchpad.net/ironic-python-agent/+bug/213854015:18
cid- https://bugs.launchpad.net/ironic/+bug/2137571 (RFE but already approved, no review necessary)15:18
TheJuliaImpressive given launchpad was down for ?2? days last week15:18
cidOh, I didn't notice. 15:19
TheJuliaThere would be 7 or 8 if it had been up all week because cardoe hit the inspection issue and I think there was another thing patches got propsoed to just fix because we couldn't file a bug15:19
TheJuliano worries :)15:19
TheJuliaOkay, So who whall be the deputy next week?15:19
cidI'm always open to it, in the absence of any other interests. I wouldn't want to have all the fun :D15:20
TheJuliaWe should really try to alternate it out15:21
TheJuliabut you do a really good job at it :)15:21
TheJuliaI'll give it a shot this week15:22
cidtks!15:22
TheJuliaSince we have no RFE's requiring review this week, onward to Open Discussion15:22
TheJulia#topic Open Discussion15:22
TheJuliaDoes anyone have anything they would like to discuss15:22
TheJulia?15:22
jandersI have a couple topics15:22
TheJuliajanders: go ahead15:23
jandersfirst - I hit this in CI an hour ago, running recheck, hopefully will clear but giving everyone a heads-up: 15:23
janders| controller | fatal: unable to look up thekelleys.org.uk (port 9418) (Temporary failure in name resolution)15:23
janderslooks like DNS failure downloading DNS package :)15:23
janders(same one dtantsur mentioned before the meeting)15:23
jandersno action required, unless we start seeing it more15:23
jandersit failed one of my patches in gate15:23
TheJuliajanders: btw, I did comment on the Redfish forum item we discussed briefly last week15:23
jandersTheJulia yeah, I saw it, ++15:24
jandersdo you expect you'll be attending tomo? we can try raise this issue for another discussion15:24
abongalecid: o/ I could take this one https://bugs.launchpad.net/python-ironicclient/+bug/2138466  and https://bugs.launchpad.net/ironic/+bug/2138129, it will be fun 15:24
jandersprob best if we're both around for the discussiobn15:25
TheJuliajanders: I will try to join tomorrow. If you want to try and raise it on the agenda then I think it might make sense. The gap really needs to be bridged somehow15:25
janders++15:25
jandersthanks TheJulia15:25
TheJuliaAnd we're sort of in a unique position there15:25
jandersyeah15:25
jandersit feels we do have a valued voice; lets use it15:25
TheJulia++15:25
cidabongale. That would be nice. o/15:26
jandersmy second topic: I would appreciate reviews on https://review.opendev.org/c/openstack/ironic/+/971152/6 and https://review.opendev.org/c/openstack/ironic/+/971150/6 - last two patches in the big firmware upgrade chain15:26
jandersright now I am re-running CI on the initial, big patch but once that's good it would be great to tick this off15:26
jandersI marked them as weekly prio15:26
TheJuliaabongale: it might be breaking, but could make sense to change the pattern of behavior, if a field value is requested with json, to return the real field name15:27
jandersboth have 1x +215:27
kubajjI also have two things15:27
kubajjA question about the possibility of implementing a node evacuation feature - i.e. have a stand-by node, if a node goes down, a user could bring over the disks and start the node with the same parameters on the other node (would probably be mostly changes on nova)15:27
kubajjMy question regarding this  topic is - do yo u think this is even possible? Would anybody else want this?15:27
TheJuliajanders: I'll trade you if you review https://review.opendev.org/c/openstack/ironic/+/973560  and https://review.opendev.org/c/openstack/ironic/+/973413 :)15:28
TheJulia(actually, I'll try to look at those this week)15:28
dtantsurabongale, TheJulia, anything around https://bugs.launchpad.net/ironic/+bug/2138129 should really consider https://bugs.launchpad.net/ironic/+bug/204991315:29
TheJuliakubajj: oh wow. That could be a big discussion. Like user physically moves the platters?15:29
dtantsurbecause it's essentially a different angle of the same discussion15:29
janderskubajj would it be expected for the image to "just work" or is there assumption there'll be manual NIC/storage/... massaging?15:30
TheJuliadtantsur: hmm, likely yes15:30
kubajjTheJulia: yep, they operate mission critical nodes, like cryogenics, and stuff15:30
dtantsurSo very like an RFE that needs a spec15:30
TheJuliaYeah, a spec most likely15:30
janderskubajj first thing that came to mind is it's borderline touching baremetal snapshots15:30
TheJuliaThe... big thing which comes to mind is NVRAM records and as previously mentioned network configuration15:31
jandersalthough I do get the difference15:31
kubajjjanders: we don't have those, right?15:31
TheJuliawe don't really have a good way to handle that15:31
janderskubajj I don't believe so15:31
TheJuliakubajj: we've talked about it, a lot over the years. (That being snapshots...) but the real issue is the complexity versues the variety of lower level interfaces15:32
TheJuliaI *do* think it would be relatively easy for us to have a means to move VIFs and ironic side configuration over15:32
kubajjjanders: yes, ideally, it would "just work" but a good starting point would be just something which would "kinda work"15:32
TheJuliaI think a spec would be a really good starting point to enumerate the requirements15:33
janderskubajj 100% agreed15:33
kubajjTheJulia: undestandable, it does sound complex15:33
kubajjTheJulia: ok, we can try to come up with one15:33
TheJuliaits almost a "move x node's config to y node"15:33
jandersjust throwing it out there: with devices in general, where I have mission critical stuff I just deploy more than I need to15:33
TheJuliaand then go move the disks in order15:33
kubajjTheJulia: yep, exactly15:33
jandersso if some die, L1 can come and do something simple-enough to keep the lights on15:34
janderssince hard work has been done preemptively 15:34
TheJuliayeah15:34
jandersnow I wonder how this maps into this concept15:34
janderscurious about any quick thoughts, otherwise happy to take this to the spec discussion15:34
jandersTLDR is - what if for some reason moving the disk isn't enough15:34
kubajjjanders: yeah, for some stuff, we do have LBs, but I do not know what exactly is the plan here, was just asked, if something like this is possible15:34
janderskubajj makes sense15:35
jandersit's a clever idea TBH15:35
janderslightweight cold spare15:35
janderskinda15:35
kubajjjanders: afaik, this is just what they do now, we are just overtaking their mini-datacentre (~600 nodes) with Ironic and they want all procedures to work15:36
kubajjwe can explore other ways of handling this later15:36
TheJuliaSo15:37
kubajjI will discuss it with them and we can come up with a spec15:37
jandersit's hard to dispute some of those battle-learned-and-tested tricks from veteran ops folks15:37
TheJuliaOne of the vendors did propose something like this as a value add years ago, but it was always using the backend SAN15:37
TheJuliaBut yeah, move the disks and if it is setup for dhcp \o/15:37
TheJuliaAlthough, if you want to force network reconfiguration... that is a bit more work15:37
TheJuliaand could require the agent to going and nuking cloud-init and then metadata would need to be updated. Regardless, there is a very interesting idea here15:38
janders++15:38
jandersI wonder about BMC config implications15:38
janders(network and other)15:38
kubajjTheJulia: I think we do not need network reconfiguration yet (still didn't manage to convince our networking team to let us touch their switches)15:38
TheJuliaWell, you evacuate to another node, so the BMC would absolutely change15:39
TheJuliajust not the config for the BMC itself15:39
TheJuliaThe way evacuate works is I think you functionally say "take this, put on this"15:39
kubajjwhat would be interesting is if they had different bios settings, would complicate it even more šŸ‘€15:39
TheJuliakubajj: well, I was more meaning workload side15:39
TheJuliaYeah, the real question is what is the resolution of the configuration changes15:40
jandersbtw none of my comments are to hint it's a bad, idea, im just thinking out loud about painpoints15:40
TheJuliaditto15:40
TheJuliaDoes anyone else have anything else to raise/discuss/plot?15:40
kubajjjanders: that's what I wanted to hear, if it was straightaway unimplementable, I would not want to start doing stuff :D15:41
kubajjThe second thing was that I wanted to thank mostepha[m] for his work upstream, he is heading back to university next week. It was great to have him around, especially for his work on the IPA15:41
TheJuliaThank you mostepha[m]!15:41
cidThank you mostepha[m]15:42
jandersThank you mostepha[m]!15:42
TheJuliaAnything else, anyone?15:43
TheJulia#topic Who is going to run our next meeting?15:45
TheJuliaWould anyone like to volunteer?15:45
TheJuliaI guess I'll run the next meeting15:46
TheJuliaOkay folks, thanks everyone! Have a great week!15:47
TheJulia#endmeeting15:47
opendevmeetMeeting ended Mon Jan 19 15:47:11 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:47
opendevmeetMinutes:        https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-01-19-15.03.html15:47
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-01-19-15.03.txt15:47
opendevmeetLog:            https://meetings.opendev.org/meetings/ironic/2026/ironic.2026-01-19-15.03.log.html15:47
jandersthanks everyone15:47
TheJuliacardoe: so, vxlan! I ran into fun with OVN on saturday where it looks like port binding never completes because OVN thinks the ports need to be shutdown because $something needs to be done by the local ovn agent, that it doesn't know/comprehend15:48
TheJuliacardoe: I'm curious, if your $other mech driver might be completing the L2 binding?15:48
cardoeAh it must be.15:49
TheJuliaOr, if we should be expecting OVN to do so, specifically it has me wondering if I have an incorrect configuration OR if I need to begin to craft another mech driver15:50
TheJuliaFWIW, I don't see OVN do anything regarding br-ex or the otherwise defined bridge15:50
TheJuliabut again, it could entirely be my config or a area where we need to highlight "correct" configuration15:50
TheJuliaFWIW, I did find a couple minorish items on the posted mech driver code and in the NGS patches. The NGS patches do the proper binding, the mech driver with minor fixes when loading *does* allocate the segment15:51
TheJuliaCode in ironic presently stalls on the port binding process because of the outstanding L2 binding record, because part of the code in OVN gets trigger to think that it needs to apply L2 binding for the local agent, I've got a workaround identified but it likely needs to live in the other mech driver if its not a misconfiguration15:55
opendevreviewJacob Anders proposed openstack/ironic master: Add hardware health monitoring via management interface  https://review.opendev.org/c/openstack/ironic/+/96694616:11
TheJuliacardoe: I guess what is confusing to me is you stressed the tagging on the physical bridge meaning the ovn/ovs side of the attachment is happening automatically when that doesn't really seem to be the case, but I could totally see a mech driver maybe doing a little bit more to make it happen. That or I have a persistent misconfiguration I've not discovered16:24
opendevreviewJacob Anders proposed openstack/python-ironicclient master: Add support for node health status field  https://review.opendev.org/c/openstack/python-ironicclient/+/96705516:28
janderscardoe working through conditional reboot patch. It seems you and Dmitry disagree on this point: https://review.opendev.org/c/openstack/ironic/+/966344/comment/69bb04f3_b189dc3b/ - first wanted to confirm you're against constants here?16:42
janders(if so we can work out a solution while I work on the follow up)16:42
cardoeI want constants19:05
janderscardoe TY, I thought that may be the case19:06
cardoeI was fixing another bug last week where a string was slightly different between files.19:08
janders++19:09
jandersfolks, I think we have a genuine CI issue. Nothing to do with the dnsmasq package this time, "missing" (pip I think) packages this time19:09
jandersit's blocking my approved change https://review.opendev.org/c/openstack/ironic/+/966344 from merging but is also affecting another unrelated change https://review.opendev.org/c/openstack/openstacksdk/+/973264 in a similar way 19:10
janderslogs:19:10
jandershttps://zuul.opendev.org/t/openstack/build/1ac3a2d3cf9245aab76a5a8c1feb51cc : No package matching 'mysql-client' is available19:10
jandershttps://zuul.opendev.org/t/openstack/build/ae483aebb4ae4b719459e8c6e5b53768: No package matching 'libpcre3-dev' is available19:11
jandersto rule out some weird problem created by the environment on my laptop, has anyone managed to get anything pass CI in the last few hours?19:11
jandersI would appreciate any pointers. 966344 is critical for a downstream bugfix19:12
janders(to be specific it became critical over the weekend)19:13
jandersIf I were to guess I suspect a dodge ubuntu repo update (or package)19:14
opendevreviewJulia Kreger proposed openstack/networking-baremetal master: WIP: l2vni baremetal mech driver  https://review.opendev.org/c/openstack/networking-baremetal/+/97388919:24
TheJuliafigures, the base OSes have likely changed and broken the plugin/bindep runs19:24
TheJuliacardoe: ^^^ first pass at moving the ml2 plugin to networking-baremetal19:25
TheJuliawith some extra logging a known workaround (for now)19:26
opendevreviewJulia Kreger proposed openstack/networking-baremetal master: WIP: l2vni baremetal mech driver  https://review.opendev.org/c/openstack/networking-baremetal/+/97388922:34
TheJuliacardoe: I'll be giving ^^^ a spin tomorrow, but I suspect that might handle the local port side as well in the model, which leaves a "bind a subport" model on the local attached switch to be a thing which is still needed22:49
opendevreviewMerged openstack/bifrost master: CI: Make bifrost-upgrade-keystone-centos-10 voting  https://review.opendev.org/c/openstack/bifrost/+/97346322:58

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