| *** jroll03 is now known as jroll0 | 07:39 | |
| rpittau | good morning ironic! o/ | 07:53 |
|---|---|---|
| rpittau | cardoe: just approved it | 08:18 |
| *** tkajinam is now known as Guest8639 | 09:04 | |
| rpittau | urgent revert! https://review.opendev.org/c/openstack/ironic-python-agent/+/941087 | 11:05 |
| rpittau | the change landed in the wrong branch | 11:05 |
| dtantsur | +W | 13:31 |
| *** milan is now known as Guest8655 | 13:48 | |
| opendevreview | Merged openstack/ironic-python-agent stable/2023.2: Fix RAID volume name https://review.opendev.org/c/openstack/ironic-python-agent/+/939960 | 13:55 |
| TheJulia | Good morning | 14:12 |
| TheJulia | Well, I think I now see why our CI fails with cirros workloads at times... :( | 14:23 |
| TheJulia | It doesn't grok to reduce the MTU when it is pinned | 14:24 |
| TheJulia | Feb 07 22:19:32 np0039796101 kernel: ovs-node-1i1: dropped over-mtu packet: 1340 > 1294 <-- once the workload boots | 14:24 |
| dtantsur | this_is_fine.jpg | 14:26 |
| rpittau | great | 14:30 |
| TheJulia | ... I've already wanted to weed out cirros from our CI, I guess this is more... "encouragement" | 14:31 |
| dtantsur | what other minimalistic stuff do we have? alpine? something-something gentoo? | 14:32 |
| dtantsur | back to torturing centos images like in the metalsmith CI? :D | 14:32 |
| * dtantsur ducks | 14:32 | |
| TheJulia | it might just be faster to download something larger | 14:32 |
| TheJulia | since public centos images also just work now | 14:32 |
| TheJulia | dunno | 14:32 |
| dtantsur | yeah, but we need partition images too | 14:33 |
| dtantsur | (that was the reason for the magic in the metalsmith CI) | 14:33 |
| TheJulia | no we don't... we already agreed to cease testing of it | 14:33 |
| dtantsur | not entirely? | 14:33 |
| dtantsur | We gave up in BIOS/grub, but I thought we were still testing UEFI asset copying? | 14:33 |
| dtantsur | If we don't, I'm sorry, I should have prevented that | 14:33 |
| TheJulia | I'd have to dig at this point | 14:34 |
| dtantsur | Metalsmith CI was testing BIOS+partition images, and we just said "okay, people should move away from that" | 14:34 |
| TheJulia | honestly, its all a blur | 14:34 |
| dtantsur | Our normal CI only tested partition images coupled with the UEFI asset copying, I believe | 14:34 |
| TheJulia | I'm 85% sure we actually removed metalsmith partition image testing | 14:35 |
| dtantsur | yep, we definitely did | 14:35 |
| TheJulia | There is the whole other aspect, the way we assemble a partition image in CI today is known to be fragile | 14:35 |
| TheJulia | although | 14:35 |
| TheJulia | it has been far better behaved as of recent | 14:35 |
| dtantsur | what isn't fragile about our CI? :D | 14:35 |
| TheJulia | touche | 14:35 |
| TheJulia | but step 0 I think at the moment is cirros | 14:36 |
| dtantsur | alpine has *some* images https://www.alpinelinux.org/cloud/ | 14:37 |
| dtantsur | claim to support uefi and cloud-init | 14:37 |
| TheJulia | hmmmm | 14:39 |
| TheJulia | yeah | 14:39 |
| TheJulia | 191 MB | 14:39 |
| TheJulia | oh, you know what | 14:40 |
| TheJulia | we only see this on the ovn job | 14:40 |
| TheJulia | ... what if ovn is not getting/sending the value | 14:40 |
| TheJulia | we don't, afaik, see this on other jobs | 14:41 |
| dtantsur | that's crazy.. but possible | 14:41 |
| TheJulia | you know what, I remember something about ovn and mtus recently | 14:42 |
| * TheJulia makes breakfast before digging more | 14:45 | |
| dtantsur | never a good idea to debug on empty stomach | 14:45 |
| * dtantsur pours more decaf in a hope to trick his brain into working | 14:46 | |
| TheJulia | heh | 14:50 |
| TheJulia | okay, it only does it when icrros is running | 14:53 |
| TheJulia | not another OS | 14:53 |
| JayF | dtantsur: TheJulia: I have gentoo support for dib in a decent place if you'd like me to try there. I do think we'd be better off using an off-the-shelf distribution if one exists though that meets our requirements. | 14:56 |
| cardoe | o/ | 15:00 |
| TheJulia | okay, I think I see what is going on | 15:01 |
| rpittau | #startmeeting ironic | 15:01 |
| opendevmeet | Meeting started Mon Feb 10 15:01:19 2025 UTC and is due to finish in 60 minutes. The chair is rpittau. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:01 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:01 |
| opendevmeet | The meeting name has been set to 'ironic' | 15:01 |
| rpittau | Hello everyone! | 15:01 |
| rpittau | Welcome to our weekly meeting! | 15:01 |
| rpittau | The meeting agenda can be found here: | 15:01 |
| rpittau | https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_February_10.2C_2025 | 15:01 |
| dtantsur | o/ | 15:01 |
| TheJulia | o/ | 15:02 |
| JayF | o/ | 15:02 |
| rpittau | this should be a quick one | 15:02 |
| rpittau | #topic Announcements/Reminders | 15:02 |
| rpittau | usual reminders! | 15:03 |
| rpittau | Standing reminder to review patches tagged ironic-week-prio and to hashtag any patches ready for review with ironic-week-prio: | 15:03 |
| rpittau | #link https://tinyurl.com/ironic-weekly-prio-dash | 15:03 |
| TheJulia | I see what is going on with mtu and all. we end up pinning the mtu on the bare interface to 1294 but ovn thinks the mtu is 1372 :( | 15:03 |
| cid | o/ | 15:04 |
| rpittau | so ovn does not know how to count? | 15:04 |
| TheJulia | full kernels likely have PMTU discovery | 15:04 |
| TheJulia | Anyway, meeting | 15:04 |
| rpittau | 2025.1 Epoxy Release Schedule | 15:04 |
| rpittau | #link https://releases.openstack.org/epoxy/schedule.html | 15:04 |
| rpittau | we're at -7 ! | 15:05 |
| rpittau | be aware that this week is oslo FF | 15:05 |
| rpittau | Flamingo PTG will take place place April 7-11, 2025! | 15:05 |
| rpittau | I'll add details and etherpad in the next days so we can start the conversation | 15:05 |
| rpittau | PTL nomination period has started! | 15:06 |
| rpittau | haven't seen any candidate so far | 15:06 |
| rpittau | #topic Discussions | 15:07 |
| TheJulia | Did we have consensus on doing DPL? | 15:07 |
| rpittau | we don't have any planned topic for today | 15:08 |
| rpittau | ok I guess we can discuss that :) | 15:08 |
| rpittau | I'm of course ok with doing DPL | 15:08 |
| JayF | If that's the route we want to go, I will volunteer to be the security liaison. It's kind of a natural extension of my work on the vmt. | 15:08 |
| dtantsur | I can probably take some role but I don't remember what roles exist :D (which may make you think that I'm not very suitable) | 15:09 |
| rpittau | and I will continue my work as release liaison :) | 15:09 |
| JayF | The one thing we'll have to nail down is that the rules now require you to have a TC liaison, someone who is on the TC who's willing to also be down as a DPL number | 15:09 |
| TheJulia | dtantsur: I think that is a perfectly acceptable indicator that maybe it is not well understood && also okay that we all forget things as time moves on | 15:09 |
| JayF | Essentially, their job is primarily auditing to make sure that we exist and are doing a job | 15:10 |
| dtantsur | It's an odd requirement, do we have the same for non-DPL project? | 15:10 |
| dtantsur | I mean, it's odd if it's limited to DPL | 15:10 |
| rpittau | I think for non-DPl is the PTL | 15:10 |
| JayF | No, it was added during my tenure on the TC as an effort to prevent DPL from being used to prop up inactive projects. | 15:11 |
| dtantsur | Well, that's not fair at all. So DPL projects are still treated as 2nd class ones. | 15:11 |
| JayF | The other thing that's different about DPL in the last year or so is that it does get reset every cycle so we'll have to opt back in. Again, just to make sure everyone who is a liaison is engaged. | 15:11 |
| dtantsur | rpittau: PTL is not a TC member | 15:12 |
| JayF | Trust me when I say that that change in policy reflected a necessary pragmatic need, and was not intended to be any kind of bucketing of one or the other as second class. | 15:12 |
| dtantsur | Well, it did exactly that | 15:13 |
| masghar | o/ What if a project does not have someone on the technical committee? | 15:13 |
| JayF | If you have a problem with the way it works now, we can choose to not use it until after you advocate and get a new policy passed. | 15:13 |
| dtantsur | Also, I trust you, but also I can easily remember many projects that went MIA with a PTL appointed and so far none that disbanded in the DPL status | 15:13 |
| rpittau | yep, I know, I skipped the part where it says that needs to be on the TC :) | 15:13 |
| JayF | masghar: in the case of ironic, I would ask a TC member. I trust to be down as a liaison | 15:13 |
| dtantsur | JayF: putting the burden on me after a questionable decision by the TC is.. questionable :) | 15:13 |
| TheJulia | ... I think my wife just suggested I should be the PTL. Please convince me not to... | 15:14 |
| JayF | Realistically, as far as the charter goes the TC already has PTL power on every project. | 15:14 |
| masghar | Oh I see, so you can just ask a TC member in case you dont have a core on the TC? | 15:14 |
| rpittau | TheJulia: don't do that! :D | 15:14 |
| JayF | dtantsur: You're talking to someone who voted yes on the policy. I don't agree at all that it's questionable. | 15:14 |
| TheJulia | rpittau: an 8th time for a corgi charm?! | 15:14 |
| dtantsur | Well, I've just explained why it's bad | 15:15 |
| JayF | dtantsur: I'll also note that during my tenure on the TC, we did tear down a lot of those projects that were being propped up by an AFK PTL | 15:15 |
| dtantsur | You're making DPL-led projects guilty until proven otherwise, contrary to PTL-led projects | 15:15 |
| cardoe | So I'm being quiet but I am a TC member. :-D | 15:15 |
| JayF | Ptl has the natural catch of an election | 15:15 |
| dtantsur | Which is not just about our feelings: this impression will probably spread further | 15:15 |
| cardoe | The goal is like Jay said to not let inactive projects just say we're using DPL. | 15:15 |
| rpittau | cardoe: I was going to tell :D | 15:15 |
| JayF | So if someone is absent for 6 months, they can be voted out. DPL is by lazy consensus | 15:16 |
| cardoe | It's a full reboot of DPL. | 15:16 |
| dtantsur | resulting in "Oh, Ironic is DPL now, it means it's going to be discontinued" | 15:16 |
| TheJulia | cardoe for president?! | 15:16 |
| dtantsur | cardoe: and what prevents them from having a fake PTL? | 15:16 |
| JayF | dtantsur: we're trying to suss out the projects that aren't active so that impression will go away. Because for the most part it was true before | 15:16 |
| cardoe | I've already discussed Ironic will look at DPL and if we don't treat it like a bad thing then Ironic will adopt it. | 15:16 |
| cardoe | dtantsur: nothing but we can replace the PTL easily and there's policy on it. | 15:16 |
| rpittau | ironicptlbot ? | 15:16 |
| TheJulia | rpittau++ | 15:17 |
| cardoe | And there's an obvious case of this person isn't doing stuff. | 15:17 |
| * dtantsur is not ready to dive too deep into yet another political issue | 15:17 | |
| dtantsur | whatever | 15:17 |
| cardoe | While in the past DPL led to a project just saying "oh need person X and they're on PTO or didn't have access to email" | 15:17 |
| dtantsur | If I can do something useful as part of the DPL model, let me kno0w | 15:17 |
| cardoe | And it went round and round. | 15:17 |
| rpittau | dtantsur: I understand your concerns and I kind of agree with them, but without a PTL we don't have a different way right now if not going as DPL with the current policy :/ | 15:18 |
| rpittau | cardoe: I guess you could be the TC liaison? | 15:18 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Improve check_vm_connectivity https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941054 | 15:19 |
| dtantsur | (we all are going to regret if you guilt-trip TheJulia or me to apply for PTL again, I promise :D) | 15:19 |
| JayF | I would run for PTL myself in this case, but I intend on taking 6 to 8 weeks off next cycle and don't think it's a good idea for me to take on that responsibility when I know I'm going to be gone for up to a third of the cycle | 15:19 |
| * TheJulia gets out the iron fist of progress | 15:19 | |
| rpittau | dtantsur: no, sorry, that's not my intention :) | 15:19 |
| JayF | I don't think it's a negative for us to go DPL, I wouldn't have suggested it if I thought it was. I think it would be reflective of how our project works in practice. | 15:19 |
| dtantsur | Yep, I fully agree with that. I got immediately scared of perception issues once you mentioned the requirement of having a supervisor to make sure we don't all run away. | 15:20 |
| JayF | Honestly, I think of it as much as a person who runs the paperwork every year to reset the DPL cycle since our automation isn't there yet | 15:20 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Improve check_vm_connectivity https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941054 | 15:21 |
| rpittau | we still have some time before deciding, do we want to think more about it? | 15:21 |
| rpittau | one more week? | 15:21 |
| dtantsur | Let's make sure that "DPL is a way for projects to die" perception is really-really gone. That is my only concern. | 15:22 |
| cardoe | That's the goal. | 15:22 |
| rpittau | ok :) | 15:22 |
| JayF | If everyone is okay with it, I can send an email out to that thread describing the liaisons that we can get volunteers for them. | 15:22 |
| rpittau | JayF: sounds great | 15:22 |
| JayF | dtantsur: I think you see it in reverse, ironic taking on DPL will remove any vestiges of that perception | 15:22 |
| dtantsur | Fair point | 15:22 |
| dtantsur | JayF: aren't the role written somewhere already? | 15:22 |
| JayF | dtantsur: ironic is one of the most active projects in all of openstack, we're unlikely to get mistaken for Some random inactive thing that gets three commits a cycle | 15:23 |
| dtantsur | lol, true | 15:23 |
| JayF | dtantsur: it is, I'm just going to link it in the thread and volunteer on the list to be the security liaison | 15:23 |
| dtantsur | JayF++ | 15:23 |
| rpittau | perfect JayF | 15:23 |
| rpittau | anything else on the DPl/PTL topic for now? | 15:24 |
| rpittau | (besides why I always type DPl with a lowercase L) | 15:24 |
| dtantsur | In the current climate, "D" must stand for "Deep". I'm not making the rules. | 15:25 |
| rpittau | :D | 15:25 |
| rpittau | ok, ok, anything else to discuss today? | 15:25 |
| dtantsur | ChatGPT for PTL? | 15:26 |
| dtantsur | sorry, sorry, I shut up already | 15:26 |
| rpittau | DeepSeek ? ;) | 15:26 |
| dtantsur | D for Deep, you see? | 15:26 |
| cardoe | Didn't someone volunteer to write a Hardware Manager doc for IPA? | 15:26 |
| rpittau | gosh! | 15:26 |
| rpittau | cardoe: I vaguely remember | 15:27 |
| JayF | cardoe: if youll write, a bug about what needs to be documented I can make time to update the documentation, I should understand it well enough | 15:27 |
| dtantsur | There are *some* docs. I guess you mean much better docs? | 15:27 |
| JayF | If someone volunteered to do it it's possibly me and I just forgot anyway | 15:27 |
| JayF | But like dtantsur references, it's already extremely well documented, so I'm curious if it's something that's just written down in a hard to find place | 15:27 |
| cardoe | Well just from reviewing Julia's bootc stuff. I thought there was some plan on documenting custom agent more and custom hardware managers more. | 15:27 |
| JayF | That'll be extremely valuable for me downstream as my MLH fellow starts today and he'll be working on hardware manager bits | 15:27 |
| cardoe | I do want to schedule something to discuss the inspector hooks and inspector rules in more detail with folks that are interested in it. | 15:28 |
| cardoe | I've got a use case that I've roughly sketched up that I'll add more details to | 15:29 |
| JayF | I'd suggest you schedule something directly with CID and I can try to make it if it overlaps with my TZ | 15:29 |
| * dtantsur registers his interest | 15:30 | |
| kubajj | o/ | 15:30 |
| cid | ++, as long as that day is not tomorrow ;) | 15:30 |
| JayF | I have something else for general discussion if there's time? | 15:30 |
| rpittau | cardoe, cid, maybe right down times/dates and discussion topics in an etherpad? | 15:31 |
| rpittau | JayF: please go ahead | 15:31 |
| JayF | Just curious if we want to make an Ironic meetup/BoF/etc for the OIF Days colocated with SCALE this year. | 15:31 |
| rpittau | errr s/right/write | 15:31 |
| JayF | Figured this would be a good place to see if we have any number of regular contributors going there. | 15:31 |
| dtantsur | JayF: where is that? | 15:31 |
| dtantsur | and when? | 15:31 |
| rpittau | was goingt o ask the same | 15:31 |
| * dtantsur is, as usually nowdays, clueless | 15:31 | |
| masghar | (also interested in the inspection hooks and rules) | 15:31 |
| JayF | Southern California USA in March; it's the North American OpenInfra Days colocated there | 15:31 |
| JayF | I'm speaking on OpenStack in the Real World | 15:32 |
| rpittau | too close and too far at the same time :D | 15:32 |
| dtantsur | Ah, USA. Pass from me. | 15:32 |
| JayF | I don't blame you; I'm not trying to say folks should come just taking an inventory :) | 15:32 |
| dtantsur | I'm very looking forward towards learning if/when/where the OIF event in Europe will be | 15:32 |
| dtantsur | (please don't tell me everyone already knows that) | 15:32 |
| JayF | I would not expect me to attend too many conferences outside of SCALE this year; maybe something in the last half of the year and maybe pycon but I'm trying to travel less this year | 15:33 |
| rpittau | dtantsur: May 22 in Paris, for me at least :) | 15:33 |
| dtantsur | rpittau: is it the same thing as last year? I cannot find anything for 2025.. | 15:34 |
| rpittau | oh it's 2024, nvm | 15:35 |
| priteau | dtantsur: There was a message just a few days ago from Helena last week on our Scientific SIG Slack. | 15:35 |
| priteau | > We are still working through the venue proposals, but we should have a decision the week after next. Timing for the Summit: we are looking at July, Aug or September, depending on the proposal we go with | 15:35 |
| opendevreview | Merged openstack/ironic-python-agent master: Revert "Configure ipa bugfix 10.1" https://review.opendev.org/c/openstack/ironic-python-agent/+/941087 | 15:35 |
| rpittau | thanks priteau | 15:35 |
| dtantsur | Exciting! | 15:35 |
| dtantsur | (summer season is pretty unfortunate, of course) | 15:36 |
| priteau | I hope they don't pick the middle of the summer when many people in Europe are on break | 15:36 |
| JayF | So I take the lack of takers as a comment that it's only gonig to be me (+ maybe TheJulia?) at SCALE for Ironic this year? | 15:36 |
| dtantsur | yeah, me for example | 15:36 |
| JayF | If so I still might form a BoF for users, but that'll be good to know | 15:36 |
| rpittau | yep summer is out for me too, especially aug-sep :/ | 15:37 |
| rpittau | unless it's in Sardegna :D | 15:37 |
| dtantsur | one can dream :D | 15:38 |
| rpittau | alright dear ironicers, any more discussion topics for today ? | 15:38 |
| TheJulia | I'll be at SCALE | 15:38 |
| TheJulia | Unknown if I'll be able to exit and reenter the US at this point | 15:39 |
| TheJulia | so it may be my only thing of the year | 15:40 |
| rpittau | :( | 15:40 |
| dtantsur | Sorry to hear about this and.. everything | 15:40 |
| TheJulia | c'est la vie... I think | 15:40 |
| cardoe | SCALE always happens when my kids are on spring break so I've not been able to make it since school for them started. | 15:41 |
| TheJulia | cardoe: I know at least *one* booth will have coloring books! | 15:41 |
| cardoe | oh if only that worked... | 15:41 |
| TheJulia | :( | 15:41 |
| TheJulia | oh well | 15:41 |
| cardoe | You've not been called into the office because your kid emailed out a password list for the entire grade in 3rd? or 4th grade... | 15:42 |
| rpittau | ok, going to close for real now, we can keep discussing after | 15:42 |
| rpittau | thanks everyone! | 15:42 |
| rpittau | #endmeeting | 15:42 |
| opendevmeet | Meeting ended Mon Feb 10 15:42:27 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:42 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-10-15.01.html | 15:42 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-10-15.01.txt | 15:42 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-02-10-15.01.log.html | 15:42 |
| TheJulia | cardoe: excellent! They will have a bright future! | 15:42 |
| cardoe | I told the school district that he provided free white hat services for them. They should learn to not generate Google passwords for all the kids using a formula that a 3rd/4th grader could figure out. | 15:43 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Improve check_vm_connectivity https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941054 | 15:43 |
| cardoe | So something for PTL... config options that aren't consistent. | 15:44 |
| cardoe | https://review.opendev.org/c/openstack/ironic/+/940333 for example that one. I tried to move the hooks config to what I felt was more consistent. | 15:44 |
| TheJulia | cardoe: could be worse... I know of a university hospital who set all passwords to "<first_initial><last_initial>abc123" | 15:45 |
| dtantsur | I agree that it's a mess | 15:45 |
| dtantsur | We're sometimes so wildly inconsistent in especially [group names] that we could be better of without groups.. | 15:46 |
| dtantsur | For instance, [DEFAULT], [conductor], [agent] and even [deploy] are used in cases where one could argue for any (or none) of these | 15:47 |
| TheJulia | Yeah, there is whole swath which likely should just get moved to DEFAULT | 15:48 |
| JayF | +++++++++++++ | 15:48 |
| dtantsur | There is one thing that is semi-consistent in Ironic, and that's what I'm referring to, cardoe: when we have some generic options and then driver-specific deviations, we tend to have the generic one somewhere logic specific ([pxe], [inspector]), and the deviating ones in driver groups ([redfish], [irmc]) | 15:48 |
| JayF | I especially hate that we have, in some places, stuff like [agent]/agent_foo_bar | 15:49 |
| dtantsur | The prior art includes, for example, kernel_append_params | 15:49 |
| cardoe | Another one I was thinking of was the cleaning steps TheJulia was helping keekz with and they were called deploy_steps. | 15:50 |
| dtantsur | ...... | 15:50 |
| TheJulia | .... that seems wrong | 15:50 |
| TheJulia | oh | 15:50 |
| TheJulia | the overrides? | 15:50 |
| cardoe | yeah they were in different sections. | 15:50 |
| * TheJulia blinks | 15:50 | |
| TheJulia | Honestly | 15:51 |
| TheJulia | there is likely the case to just let someone propose an appropriate shuffling | 15:51 |
| TheJulia | because anything else might get us into the territory of trying to design perfection | 15:51 |
| dtantsur | This sounds right but also like a huge bikeshedding exercise :) | 15:51 |
| TheJulia | (yeah, see, trying to avoid building a shed and having to paint it... ) | 15:52 |
| TheJulia | ;) | 15:52 |
| cardoe | yeah I'll go back to my corner and close my mouth. | 15:52 |
| * TheJulia assumes someone has a bike shed we can just use a crane to deliver | 15:52 | |
| TheJulia | cardoe: no no! | 15:52 |
| cardoe | I realize now it's like throwing a holy hand grenade out there. | 15:52 |
| dtantsur | My personal regret: "direct" deploy. If I had known that iscsi would be gone rather soon.... | 15:52 |
| TheJulia | dtantsur: perfection is the enemy of good | 15:53 |
| TheJulia | Anyway, cardoe, please propose something that makes more sense to you | 15:53 |
| JayF | It's still the agent driver in my heart 😂 | 15:53 |
| cardoe | Okay will do. | 15:53 |
| dtantsur | cardoe: I definitely don't want to discourage you to think about organizing the configuration options better. I'm merely pointing out that there is little existing logic you can rely on. | 15:53 |
| TheJulia | your working at an operator, it brings more value that way becuase your using a different set of lenses | 15:53 |
| dtantsur | JayF: exactly, nobody is using the word "direct" | 15:54 |
| TheJulia | most of us, at least on the pure dev side get much more hyper-focused into specific areas and it is difficult to step back and look at the big picture at times | 15:54 |
| TheJulia | We totally need to do that, but difficult is that super fun word :) | 15:54 |
| cardoe | I've just had to make a glossary to Ironic terms if you will (I've got the same for Neutron) | 15:54 |
| dtantsur | Glossaries are great, all docs should have them | 15:55 |
| cardoe | And it just feels like either I should figure out a way how to upstream this for others to find. | 15:55 |
| TheJulia | ++ | 15:55 |
| JayF | Or we look back at the big picture, have some version of this giant conversation in our own heads and return to something more achievable 😂 | 15:55 |
| JayF | Maybe that's just me. | 15:55 |
| cardoe | Well like the "direct" thing you mention. I've got something written up about deploy interfaces and specifically it starts "the direct interface is not direct" | 15:55 |
| dtantsur | cardoe: it may be as simple as slapping the glossary page on the top of our documentation tree and add links to where each concept is better describedd | 15:55 |
| dtantsur | "the direct interface is not direct" is a damning statement to my ability to name things :D | 15:55 |
| JayF | It's direct from a lens where a driver exists where you don't write the image directly onto the disk. That entire perspective is basically gone now though. | 15:56 |
| cardoe | dtantsur: hey don't worry. At many jobs it was a running job that I wasn't allowed to name things cause my names stuck and they were awful. | 15:56 |
| dtantsur | lol awesome | 15:56 |
| * dtantsur hopes few people remember ironic-discoverd | 15:57 | |
| cardoe | Ultimately what I'm thinking about in my head is minimizing my glossary / conversion doc and getting that stuff into upstream. Because I figure if our folks have to ask me about it then there's other people in the world that have had the same question. | 15:58 |
| masghar | Even a few improvements to the config structure could be helpful, after consensus. Also the glossary upstream | 16:00 |
| masghar | I, too have wondered about the direct deploy :D | 16:00 |
| JayF | dtantsur: fwiw, when I tell that story to folks like cid to explain why inspector was ever separate, usually my reasoning is something like we were too silly to realize that this was in our scope. I would never ever think of that as a failing on your part, more a failing on the project to anticipate we would want all of that scope | 16:00 |
| TheJulia | well, we just "directly" write it to disk ;) | 16:00 |
| masghar | are we more indirect in other cases 😅 | 16:01 |
| TheJulia | This all goes back to the hardest problem in computer science, naming | 16:01 |
| JayF | masghar: The default driver for ironic used to be one that operated on mounted iscai volumes on the conductor | 16:01 |
| masghar | oh, wow | 16:02 |
| JayF | masghar: when Rackspace introduced the agent driver back in 2014, it was thought it might not be practical to transmit the whole image over the network for every node. Transmit the whole image over the network for every node... I think we figured it out 😂 | 16:02 |
| masghar | I see, I see | 16:02 |
| TheJulia | The whole reason we got rid of iscsi deploy was because it was a pain to troubelshoot and it wasn't realy able to recover a short transient failure | 16:03 |
| TheJulia | so we would see cases where the network interface would bounce due to the switch, port might be down 15 seconds, and that would be enough to break the deploy | 16:04 |
| masghar | oh that doesnt sound pleasant to debug | 16:06 |
| TheJulia | okay, I see what is going on with the ovn job | 16:07 |
| TheJulia | the job is failing because we are pinning the MTU down further than the MTU neutron is running with and asserts as the across the board default | 16:08 |
| TheJulia | Neutron ends up with an mtu of 1372, and ironic bridge interface wiring is artifically then reducing that down to 1294 | 16:08 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Improve check_vm_connectivity https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941054 | 16:13 |
| cardoe | oh if anyone wants to further the pre-commit march... https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940576 and https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940577 | 16:25 |
| cardoe | is https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940588 okay as a "git revert" or do I need to do it via gerrit? | 16:25 |
| TheJulia | thats fine | 16:28 |
| rpittau | good night! o/ | 16:54 |
| opendevreview | Julia Kreger proposed openstack/ironic master: CI: Only artificially pin down the mtu in multinode https://review.opendev.org/c/openstack/ironic/+/941136 | 16:55 |
| TheJulia | I *think* that should address the rando ovn job failures | 16:56 |
| JayF | I thought there was a reason we capped the minimum at 1280 | 17:02 |
| JayF | ah, yeah, IPv6 doesn't work below that per the comment | 17:02 |
| dtantsur | yep | 17:03 |
| TheJulia | oh, heh | 17:03 |
| TheJulia | well, I guess we'll see how CI responds | 17:03 |
| cardoe | ugh.. I think I'm gonna have to make a custom deploy method for VMware ESXi... At least I recently looked at Julia's bootc stuff and that helped me understand it. | 17:18 |
| TheJulia | cardoe: most people I talk to do small lightweight esxi disk images, fwiw | 17:19 |
| cardoe | You got any reference? | 17:20 |
| cardoe | I'm over here looking at what they gave me and it's literally says "Linux system admins should find this natural" and then proceeds to throw a bunch of PowerShell non-sense at me. | 17:21 |
| JayF | sounds like every experience I had with windows engineers at your employer :P | 17:23 |
| JayF | we asked for cloud-init integration, we got a giant unreadable (power)shell script that set up windows :D | 17:23 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Improve check_vm_connectivity https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941054 | 17:25 |
| cardoe | TheJulia: even if it's a Google search term or something to throw into ChatGPT. | 17:28 |
| TheJulia | cardoe: I tried to put upstream docs up but... yeah, it was blocked | 17:29 |
| TheJulia | steve has posted new/interesting details to https://review.opendev.org/c/openstack/ironic-specs/+/938526 | 17:32 |
| TheJulia | I've +2'ed it | 17:32 |
| cardoe | Yeah I'm a mega +2 on that spec. | 17:37 |
| JayF | since I was the third +2 I landed the spec | 17:39 |
| opendevreview | Merged openstack/ironic-prometheus-exporter master: fix bashate error in script https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940576 | 17:39 |
| opendevreview | Merged openstack/ironic-prometheus-exporter master: enable pre-commit lints from main ironic repo https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940577 | 17:39 |
| opendevreview | Merged openstack/ironic-prometheus-exporter master: Revert "temp hack for failing flake8 check" https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/940588 | 17:39 |
| cardoe | What else do we need to pre-commit, JayF? | 17:42 |
| cardoe | Probably bifrost? | 17:42 |
| JayF | When I need to do a "for n in all repos" change for ironic I always check the projects.yaml in governance or the bug deputy doc page for a list | 17:43 |
| JayF | if you wanna make sure all the i's are dotted so to speak, I'd check those | 17:43 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent master: follow-up: update release note for bootable container work https://review.opendev.org/c/openstack/ironic-python-agent/+/941139 | 17:44 |
| JayF | +2A on that followup | 17:47 |
| TheJulia | Thanks! | 17:47 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: move hooks execution into shareable module https://review.opendev.org/c/openstack/ironic/+/940332 | 17:49 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: allow multiple inspection interfaces to load hooks https://review.opendev.org/c/openstack/ironic/+/940333 | 17:49 |
| opendevreview | Merged openstack/ironic-specs master: Graphical Console Support https://review.opendev.org/c/openstack/ironic-specs/+/938526 | 17:50 |
| cardoe | JayF: perfect thanks. | 17:50 |
| cardoe | TheJulia: So that shareable module thing above... I'm not sure if I love the "on_error_plugin_data" callback but I don't necessarily have any other ideas other than putting some methods on the BaseInspect class? Do you have any suggestions? | 17:51 |
| cardoe | Or honestly anyone that wants to give feedback. I'm not sure how folks feed about adding methods on those driver interfaces since it seems like the prior art is to make standalone functions. | 17:51 |
| opendevreview | Julia Kreger proposed openstack/ironic master: CI: Only artificially pin down the mtu in multinode https://review.opendev.org/c/openstack/ironic/+/941136 | 17:54 |
| TheJulia | Okay, that one should be happier in CI | 17:54 |
| opendevreview | Merged openstack/ironic-specs master: Support for bootable containers https://review.opendev.org/c/openstack/ironic-specs/+/933620 | 18:08 |
| TheJulia | cardoe: could you point me to exactly where your setting that? | 18:10 |
| cardoe | Sorry.. can't seem to get a link directly in gerrit to the line change.... "def run_inspection_hooks(task, inventory, plugin_data, hooks, on_error_plugin_data):" it's line 525 of inspect_utils.py of https://review.opendev.org/c/openstack/ironic/+/940332 | 18:11 |
| cardoe | That way "_store_logs()" stays put in the agent code. | 18:12 |
| TheJulia | ahh, I see | 18:19 |
| TheJulia | ... I dunno, I think that is the most direct way of doing that, honestly | 18:19 |
| TheJulia | rpittau: since https://review.opendev.org/c/openstack/ironic/+/940800 is docs, if you want to workflow it go right ahead | 18:31 |
| JayF | +2A on that | 18:34 |
| TheJulia | fair enough | 18:39 |
| opendevreview | Merged openstack/ironic bugfix/28.0: Configure ironic bugfix 28 https://review.opendev.org/c/openstack/ironic/+/941088 | 18:44 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: move hooks execution into shareable module https://review.opendev.org/c/openstack/ironic/+/940332 | 18:46 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: allow multiple inspection interfaces to load hooks https://review.opendev.org/c/openstack/ironic/+/940333 | 18:46 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: allow running inspection hooks on redfish interface https://review.opendev.org/c/openstack/ironic/+/933066 | 18:46 |
| opendevreview | Merged openstack/ironic master: Add basic documentation on metal3-integration job https://review.opendev.org/c/openstack/ironic/+/940800 | 18:59 |
| opendevreview | Satoshi Shirosaka proposed openstack/ironic master: Remove Ipmi from the default enabled vendor interfaces https://review.opendev.org/c/openstack/ironic/+/941153 | 19:02 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add tests to validate trunks https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 | 19:18 |
| opendevreview | Merged openstack/ironic-python-agent master: Bootable container support https://review.opendev.org/c/openstack/ironic-python-agent/+/940178 | 19:26 |
| TheJulia | woot | 19:57 |
| opendevreview | Julia Kreger proposed openstack/ironic master: bootc deploy interface - for bootable containers https://review.opendev.org/c/openstack/ironic/+/937897 | 19:58 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Validate network data for portgroups https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/940678 | 20:04 |
| JayF | Is there anyone with like, 5 minutes to spare to rubber duck something with me? | 21:29 |
| JayF | I've managed to convince myself deploy_{kernel,ramdisk}_by_arch is only hooked up for redfish and ilo virtual media boot | 21:29 |
| JayF | but it seems exceedingly unlikely we've had a bug around that feature for this long | 21:29 |
| JayF | but afaict it's only referenced in ironic.drivers.utils.get_agent_kernel_ramdisk which is only called by redfish/ilo boot.py | 21:32 |
| JayF | I feel like I have to be missing something, because otherwise basically our pxe/ipxe boot path is old untouched code and the only updated ones are redfish/ilo | 21:32 |
| TheJulia | maybe in like 5 | 21:34 |
| TheJulia | you very well may be right | 21:41 |
| JayF | yeah, I'm more and more convinced the more I look | 21:43 |
| opendevreview | Vasyl Saienko proposed openstack/ironic-tempest-plugin master: Add tests to validate trunks https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/941013 | 21:43 |
| JayF | I think there'll be a backportable bugfix | 21:43 |
| JayF | my downstream tried to backport the patch, it backported cleanly but didn't work | 21:43 |
| JayF | this is *not* what I expected to find | 21:43 |
| JayF | if you're in agreement with me, I'll dig into a fix | 21:43 |
| TheJulia | yeah, its a bug from what I can see | 21:46 |
| JayF | ack, thanks for sanity checking | 21:47 |
| JayF | when something that old is that broken it's nice to know you didn't miss anything | 21:47 |
| JayF | https://bugs.launchpad.net/ironic/+bug/2097798 has been created about this issue, marked as triaged/medium and assigned to me | 21:59 |
| TheJulia | Okay, the mtu change looks like it is good, but I'm rechecking it now | 22:14 |
| opendevreview | Michael Sherman proposed openstack/networking-generic-switch master: Fix binding l2-only ports by setting connectivity property https://review.opendev.org/c/openstack/networking-generic-switch/+/941172 | 23:52 |
| opendevreview | Verification of a change to openstack/ironic master failed: bootc deploy interface - for bootable containers https://review.opendev.org/c/openstack/ironic/+/937897 | 23:54 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!