| opendevreview | Merged openstack/ironic-ui master: Imported Translations from Zanata https://review.opendev.org/c/openstack/ironic-ui/+/1006150 | 09:23 |
|---|---|---|
| opendevreview | Merged openstack/ironic unmaintained/2024.1: fix broken project scoping https://review.opendev.org/c/openstack/ironic/+/1004706 | 09:40 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent master: Maintain plugin requirements as optional dependencies https://review.opendev.org/c/openstack/ironic-python-agent/+/1000668 | 10:52 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent master: Maintain plugin requirements as optional dependencies https://review.opendev.org/c/openstack/ironic-python-agent/+/1000668 | 10:52 |
| opendevreview | Merged openstack/ironic-python-agent master: Replace own utility to wrap IPv6 address https://review.opendev.org/c/openstack/ironic-python-agent/+/1007273 | 11:59 |
| opendevreview | Bernd Kuespert proposed openstack/python-ironicclient master: Fix filecache on Windows: use threading lock instead of fcntl https://review.opendev.org/c/openstack/python-ironicclient/+/1008017 | 12:32 |
| TheJulia | brraaains | 12:52 |
| iurygregory | good morning Ironic o/ | 12:59 |
| TheJulia | good morning | 12:59 |
| iurygregory | TheJulia, hey I will be afk for ~ 1h30min need to solve some problems... but I will be free after that o/ | 12:59 |
| TheJulia | that sounds perfect, allows Jay to have coffee, cardoe to spray water on cats climbing the racks. | 13:01 |
| TheJulia | And allows me to possibly go for a run this morning | 13:02 |
| iurygregory | ++ | 13:03 |
| JayF | I am not around on Friday mornings, like 730-930 local every week | 13:13 |
| TheJulia | yup | 13:14 |
| TheJulia | JayF: around when you tended to return is when I was thinking would be a good time because it should be while Iury is still around | 13:20 |
| TheJulia | fwiw | 13:20 |
| * cardoe walks the racks with his spray bottle out. | 13:21 | |
| TheJulia | you around still? | 13:22 |
| TheJulia | just figuring better to wait for a little sunlight before I go force march the corgi into my "i will run soon" speed | 13:22 |
| cardoe | I'm free whenever | 13:22 |
| TheJulia | It would be kind of funny to have a data center full of cats with servers. Cat's sleeping on top of racks, on top of transformers, etc. | 13:26 |
| TheJulia | PDU cabinets where all the bus bar warmth is | 13:26 |
| dtantsur | Chaotic good vibes won't hurt modern data centers | 13:28 |
| TheJulia | and could encourage the liquidcool most things model! | 13:28 |
| TheJulia | or cryocool! | 13:29 |
| dtantsur | Unless somebody enjoys fluff on the CPUs and other sensitive parts :D | 13:30 |
| TheJulia | This is part of the way to get Max Headroom. | 13:32 |
| * TheJulia hopes you've seen it | 13:32 | |
| Mahnoor | I'm wondering whether Julia has been looking at bugs these last 2 weeks? | 13:39 |
| Mahnoor | I recall signing up to bug deputy... | 13:39 |
| Mahnoor | If I see bugs that are 'In Progress' with patches from someone, I hope nobody minds me assigning the bug to that person :) | 13:53 |
| dtantsur | This is usually the right thing to do | 13:53 |
| Mahnoor | alrightie | 13:57 |
| TheJulia | I have been, but yeah, some items, espescially those which start out as private security, I may just tend to do a first pass change for discussion and see what thoughts are | 14:17 |
| iurygregory | I'm back o/ | 14:42 |
| opendevreview | Verification of a change to openstack/bifrost stable/2026.1 failed: [stable only] CI: Fall back to HTTPS when cirros 0.5.3 is not cached https://review.opendev.org/c/openstack/bifrost/+/1008072 | 14:57 |
| Mahnoor | alright, I will go through the bugs and put a list in the agenda, thanks Julia! | 14:59 |
| TheJulia | Mahnoor: some of it is already there, fyi | 15:13 |
| TheJulia | Mahnoor: so the delta in this week is the needful | 15:14 |
| cardoe | Uggla: can we get https://review.opendev.org/c/openstack/nova/+/1004890 https://review.opendev.org/c/openstack/nova/+/1004899 and https://review.opendev.org/c/openstack/nova/+/1005073 onto a review queue? Those are backports. | 15:16 |
| Uggla | cardoe, putting them in the doc. I will also advertise them in next upstream meeting. | 15:18 |
| cardoe | thank you | 15:18 |
| cardoe | Those are the backports of the nova/ironic improvements I highlighted during the OpenInfra Live recording | 15:19 |
| TheJulia | is in a little over an hour for schedule creation? | 15:25 |
| TheJulia | Jay should hopefully be back by about then or shortly their after | 15:26 |
| iurygregory | oh, so I go grab some lunch o/ | 15:26 |
| iurygregory | 12:26 here | 15:27 |
| iurygregory | brb | 15:27 |
| Mahnoor | ThJulia: ack | 15:28 |
| TheJulia | folks caring about older branches, do we need to backport https://review.opendev.org/c/openstack/ironic/+/986597 ? | 15:43 |
| Mahnoor | Added a 2nd RFE to the agenda | 15:53 |
| Mahnoor | Also found these 3 new bugs: https://bugs.launchpad.net/python-ironicclient/+bug/2168967, https://bugs.launchpad.net/sushy/+bug/2169279, https://bugs.launchpad.net/ironic/+bug/2169173 | 15:54 |
| Mahnoor | Thats all | 15:54 |
| Mahnoor | (just triaged and updated the counters for those 3 bugs) | 15:54 |
| iurygregory | TheJulia, upstream wise I would say 1 or 2 stable versions.. | 16:05 |
| TheJulia | hmm, joy | 16:31 |
| iurygregory | that is my thinking | 16:32 |
| iurygregory | downstream we will have to do a bunch of backports I think XD | 16:32 |
| TheJulia | I'll create an ad-hoc call for the schedule in a couple minutes, got distracted by supercomputing's schedule | 16:33 |
| iurygregory | ack | 16:33 |
| iurygregory | yay CSB rebooted again... | 16:36 |
| TheJulia | https://meet.google.com/gmd-qmdo-vup | 16:38 |
| iurygregory | connecting things, will join a bit | 16:40 |
| TheJulia | k | 16:41 |
| JayF | Give me 5-10 and I'll be in | 16:44 |
| JayF | https://review.opendev.org/c/openstack/oslo.utils/+/1008287 might be relevant for us | 17:35 |
| JayF | yeah nevermind re ^ | 17:52 |
| JayF | cid: Mahnoor: https://etherpad.opendev.org/p/ironic-ptg-2027.1 please look at line 56+ and make sure any architectural items you wanna discuss end up listed. We're aiming for topics that we can timebox a high level overview in 10 minutes | 18:02 |
| JayF | (looking to give you enough information to know how to get stronger in it) | 18:02 |
| JayF | clif: ^ you might have some insight, too | 18:02 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: Fix TBN Portgroup creation accessing nonexistent vendor field https://review.opendev.org/c/openstack/ironic/+/1008516 | 18:21 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: Make TBN comparisons against unset variables never match https://review.opendev.org/c/openstack/ironic/+/1008517 | 18:21 |
| cardoe | clif: ^^ | 18:21 |
| cardoe | https://review.opendev.org/c/openstack/oslo.service/+/992774 would be a good one for us with eventlet | 18:22 |
| opendevreview | Iury Gregory Melo Ferreira proposed openstack/ironic stable/2026.2: Fix Lenovo UEFI boot loop in prepare_instance flow https://review.opendev.org/c/openstack/ironic/+/1008518 | 18:23 |
| clif | cardoe: did portgroup ever have a vendor field? | 18:42 |
| cardoe | No | 18:42 |
| clif | dang | 18:43 |
| cardoe | so we're hitting baremetal port vif attach <blah> <blah> --portgroup-uuid <uuid-here> throwing a 500 error | 18:43 |
| clif | +2, sorry for the pain | 18:47 |
| JayF | Make sure we backport those fixes | 18:58 |
| cardoe | clif: don't apologize instead see it as "a thing I wrote is being used by someone other than me" | 19:10 |
| clif | true :D | 19:10 |
| TheJulia | Regarding idrac into redfish, Is there any reason to have a session, I think the feeling is "as long as we have an upgrade path handled of some sort, its likely okay? and don't need to litigate it the ptg. | 19:19 |
| TheJulia | dtantsur: ^^ | 19:19 |
| TheJulia | Okay fellow bi-pedal creatures and AI instances! We have a schedule on the PTG etherpad: https://etherpad.opendev.org/p/ironic-ptg-2027.1 Please raise any concerns or questions. I've reached out to Cinder as well to pin down the time. | 19:28 |
| cardoe | TheJulia: yeah I agree. No need to talk about it. | 19:54 |
| cardoe | Like I was saying... I wanna have that quirks mode | 19:54 |
| cardoe | That you can have a driver_info.quirks field that has some named quirks | 19:55 |
| TheJulia | if vendor == 'cardoe's magical hardware' ;) | 19:55 |
| cardoe | No no we bubble stuff up we've seen | 19:55 |
| JayF | this is sorta against the direction we're going though | 19:55 |
| JayF | we had a pluggable model that we're punting on | 19:55 |
| JayF | to a non-pluggable model | 19:55 |
| TheJulia | The model under the hood is largely "by vendor" handling the quarks, but sometimes its by model as well or even state | 19:56 |
| * TheJulia looks at Dell gear | 19:56 | |
| cardoe | The pluggable model isn't working out correctly though | 19:56 |
| JayF | and I get what you're nodding at, from a UI perspective, but I'm just noting we handle that sorta mismatch/quirk in Ironic/IPA often with split out plugins | 19:56 |
| cardoe | Cause oh wait Julia forgot to add ipxe boot to dell but only added it to hpe! | 19:56 |
| JayF | cardoe: I'm not saying I disagree with the direction, per se, just noting that we are taking 1 step back to take 2 forward :) | 19:56 |
| JayF | and if we legit had a way to mix in quirks, hooking that up to stevedore would still be cool | 19:57 |
| TheJulia | cardoe: grub all the things for secure boot. | 19:57 |
| JayF | even if vendor-aligned plugins aren't the solutino | 19:57 |
| TheJulia | OH NOES | 19:57 |
| cardoe | JayF: today we try to do a POST with an empty body to the X endpoint 3 times and then try again with {} 3 more times... but whoops vendor Y locks you out if you have 3 bad attempts. | 19:57 |
| TheJulia | has anyone looked at the ipxe signed loader artifact stuffs? | 19:57 |
| JayF | cardoe: sounds like maybe we should default to 2 (in addition to having the ability to quirk on it) | 19:57 |
| cardoe | JayF: yeah mixing in quirks with stevedore is where I would love to end up. | 19:57 |
| TheJulia | (like, it sounds like they got MSFT to sign them a shim | 19:57 |
| JayF | My downstream has never expressed interest in secure boot at all | 19:58 |
| TheJulia | cardoe: there is some amount of those sorts of cases which really need to surface back as bugs | 19:58 |
| TheJulia | cardoe: and not be treated as "downstream quarks" | 19:58 |
| cardoe | I don't disagree. | 19:59 |
| JayF | well, I mean, both? | 19:59 |
| cardoe | I'm just talking about all the existing stuff we have today. | 19:59 |
| JayF | using cardoe's example, we can both 1) define that fix as an opt-in quirk AND 2) detect the hardware and auto-quirk it | 19:59 |
| JayF | you don't want only #2, you want #1 as well so operators can apply the quirks as needed to fresh hardware | 19:59 |
| JayF | and don't rely on ironic releasing updates with awareness of new firmware versions or w/e | 20:00 |
| cardoe | the redfish driver has a ton of if/else today and when I get new gear that needs that work around I need to cut a new patch to Ironic to enable it | 20:00 |
| JayF | but instead if it was like | 20:00 |
| JayF | if vendor == 'badVendor' or node.driver_internal_info.get('quirky_badVendor_quirk_17') | 20:00 |
| JayF | you save the patch | 20:00 |
| cardoe | Yeah | 20:01 |
| JayF | and we can explictly document "please file a bug with $hardwareInfo if you have to apply a quirk" | 20:01 |
| JayF | then some day, over the rainbow, we can just have mappings of hardware:quirks and just check quirks | 20:01 |
| JayF | this is basically feature flagging | 20:01 |
| TheJulia | okay, I an see that | 20:02 |
| cardoe | I was suggesting driver_info so that if someone reports a bug we can easily get them to dump the quirks enabled for that node. | 20:02 |
| cardoe | And we can say "see the quirk you have enabled? that means bug your hardware vendor" | 20:02 |
| JayF | I was thinking more the reverse cardoe lol | 20:02 |
| JayF | "oh, you have to enable quirks X Y Z for hardware Foo? We'll do that automatically in Ironic N+1" | 20:03 |
| cardoe | or "your LLM enabled quirk_17 on hardware that doesn't need quirk_17 ya silly goose" | 20:03 |
| cardoe | oh that's a good point too. We can use that as a data point. | 20:03 |
| JayF | yeah that's what I'm saying, we'd be building a programatic way to talk about these things | 20:03 |
| JayF | some of the quirks wouldn't even have to be pure flags | 20:03 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: Add iDRAC BMC config steps (NTP, DNS, OIDC) https://review.opendev.org/c/openstack/ironic/+/1007083 | 20:04 |
| JayF | quirk_redfish_auth_session_timeout = 5000 | 20:04 |
| cardoe | btw... ^ I don't love the patch... it's not mergable either | 20:04 |
| cardoe | It spews secrets all over my logs so the real patch is also a bunch of "no no no delete... no" | 20:05 |
| TheJulia | yeah, there is a time in place to auto-redact step contents from logs, I just don't think that has been modeled/figured out. | 20:11 |
| TheJulia | {"step": "foo", "interface": "bar", "args": "{'foo': 'bar', 'secret': '******'}} | 20:12 |
| TheJulia | And then we can call conductor.utilities.inflate_weekend_bouncy_castle which checks for EOD friday as well whenever redaction requests are needed | 20:13 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!