| opendevreview | Merged openstack/ironic master: DB: Load only one instance for RPC interactions https://review.opendev.org/c/openstack/ironic/+/892397 | 00:44 |
|---|---|---|
| opendevreview | Merged openstack/ironic-python-agent master: Add additional mock tests to unit tests for read only devices. Change ordering to ensure mock tests work correctly. https://review.opendev.org/c/openstack/ironic-python-agent/+/897611 | 05:29 |
| rpittau | good morning ironic! o/ | 06:53 |
| rpittau | dtantsur: re: virtual media attaching/detaching API, I agree, I didn't see a clear path for the caching and we need to focus on the API first | 07:06 |
| dtantsur | rpittau, it's kinda the other way around: caching is a hard requirement, while username/password/insecure stuff is somewhat I invented | 07:23 |
| rpittau | mmm ok, I was not considering username/password/insecure at all as it's commented out | 07:29 |
| rpittau | now I mean | 07:30 |
| dtantsur | yeah, that's what I removed to make the code compatible with the caching code in image_utils | 07:46 |
| dtantsur | Which, in turn, needs a lot of refactoring.... | 07:46 |
| rpittau | ok, that makes sense | 08:14 |
| mallik | dtantsur, hi | 11:51 |
| mallik | partition images are still supported in Ironic? I am trying with latest master and seeing some issues with partition images. | 11:53 |
| mallik | can some can confirm whether partition images are supported? | 11:54 |
| dtantsur | I can confirm, there are supported. | 12:09 |
| dtantsur | * they are | 12:09 |
| mallik | dtantsur, thanks for the confirmation. | 12:20 |
| TheJulia | good morning | 13:01 |
| * TheJulia tries to wake up | 13:09 | |
| opendevreview | Julia Kreger proposed openstack/ironic master: CI: Fix our internal MTU settings https://review.opendev.org/c/openstack/ironic/+/893112 | 13:13 |
| TheJulia | just a rebase since the parent patch still needs review | 13:14 |
| iurygregory | good morning Ironic | 13:37 |
| TheJulia | good morning iurygregory | 13:37 |
| opendevreview | Mahnoor Asghar proposed openstack/ironic master: Add inspection hooks https://review.opendev.org/c/openstack/ironic/+/892661 | 13:39 |
| * dtantsur looks at image_utils and gets a large torch | 14:32 | |
| TheJulia | dtantsur: Oxyacetylene! | 14:35 |
| dtantsur | Sounds good, +2 | 14:35 |
| TheJulia | I ordered two pieces of 316 stainless recently to try laser cutting | 14:36 |
| TheJulia | odds are I need a better laser.... | 14:36 |
| TheJulia | It may be better for laser wending actually.... | 14:40 |
| JayF | heads up, after Ironic meeting today I'm going to be working on sorting and assigning times to PTG sessions | 14:57 |
| JayF | if anyone has input, leave it here or I can do it sync with others | 14:57 |
| JayF | my primary focus is just getting something done, we can shift times around if folks have conflicts | 14:58 |
| TheJulia | ++ | 14:58 |
| iurygregory | o/ | 15:00 |
| iurygregory | opa | 15:00 |
| iurygregory | ops* | 15:00 |
| JayF | you can start it if you want it on time :P | 15:02 |
| JayF | #startmeeting ironic | 15:02 |
| opendevmeet | Meeting started Mon Oct 9 15:02:14 2023 UTC and is due to finish in 60 minutes. The chair is JayF. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:02 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:02 |
| opendevmeet | The meeting name has been set to 'ironic' | 15:02 |
| rpittau | o/ | 15:02 |
| TheJulia | o/ | 15:02 |
| JayF | #topic Announcements/Reminder | 15:02 |
| iurygregory | o/ | 15:02 |
| JayF | #info Standing reminder to review patches tagged ironic-week-prio and to hashtag any patches ready for review with ironic-week-prio: https://tinyurl.com/ironic-weekly-prio-dash | 15:02 |
| JayF | #topic Action items from previous meeting | 15:02 |
| JayF | I was going to release a new NGS | 15:02 |
| JayF | I tried to; the changes in stable/2023.2 hadn't landed | 15:02 |
| JayF | and I got pushback from release team for trying to do a feature-version-bump | 15:03 |
| dtantsur | o/ | 15:03 |
| JayF | IMO: it's too late in the cycle to backport those features to NGS | 15:03 |
| TheJulia | I thought we agreed like everything waiting for ngs was basically bug fixes | 15:04 |
| TheJulia | or was there more? | 15:04 |
| JayF | but is there anyone invested in getting that done? Otherwise I'll ensure the bugfix patch (I think there's one?) is backported and released | 15:04 |
| JayF | there was absolutely stuff I don't feel belongs in a bugfix release bump | 15:04 |
| JayF | let me get it up\ | 15:04 |
| TheJulia | okay, we would need to take a look at it, we can sift through it later, meeting now :) | 15:04 |
| JayF | #link https://review.opendev.org/q/project:openstack/networking-generic-switch+branch:stable/2023.2+status:open | 15:04 |
| JayF | wanted to get the link into the log :) | 15:05 |
| JayF | I think the honor ngs_save was what I was thinking was a bugfix | 15:05 |
| JayF | the fake one is whatever | 15:05 |
| JayF | the last one is the one I think is a bit scary to backport | 15:05 |
| TheJulia | first and last looks like bug fixes | 15:05 |
| TheJulia | I could go along with that | 15:05 |
| JayF | ack; I'll take that action | 15:05 |
| JayF | #action JayF to backport honor ngs_save fix to stable/2023.2, cut a x.y.N+1 release of NGS | 15:06 |
| JayF | cool | 15:06 |
| JayF | #topic Bobcat Release | 15:06 |
| JayF | We had one. Thank you! | 15:06 |
| JayF | also TheJulia wrote a nice blog about service steps, you should read it | 15:06 |
| JayF | I recognized it because I was going to write a nice blog about service steps but she stole her own thunder before I could ;) | 15:07 |
| TheJulia | blame diablo_rojo | 15:07 |
| JayF | I don't think there's anything meaty to discuss here? | 15:07 |
| JayF | #topic [rpittau] Pillow/blockdiag/seqdiag issue | 15:07 |
| JayF | #link | 15:07 |
| JayF | #link https://bugs.launchpad.net/ironic/+bug/2026345 | 15:07 |
| JayF | #link https://review.opendev.org/c/openstack/requirements/+/897537 | 15:07 |
| rpittau | in short blockdiag has been discontinued since a while | 15:08 |
| JayF | Looks like we have an old/breaking library in our requirements. Someone invested in our state diagram should migrate it to graphwiz or similar. | 15:08 |
| rpittau | nad the recent Pillow update broke compatibility | 15:08 |
| TheJulia | the state machine diagram? | 15:08 |
| rpittau | I've added the topic to the PTG section of documentation | 15:08 |
| JayF | I assumed this https://docs.openstack.org/ironic/latest/user/states.html was the use of blockdiag | 15:09 |
| JayF | is that a bad assumption? | 15:09 |
| rpittau | there are multiple entries of seqdiag that need to be migrated | 15:09 |
| JayF | service is not reflected in there either, I guess | 15:09 |
| TheJulia | dunno, but took a look a while back and the rendering library was also sort of last updated on python2 | 15:09 |
| TheJulia | so we might just need to find a new tool to render it | 15:09 |
| rpittau | my proposal is to move to PlantUML which is based on graphviz | 15:09 |
| JayF | I'm happy to accept whatever the community or the person doing the work wants to use :D | 15:10 |
| TheJulia | ++ | 15:10 |
| rpittau | #link https://plantuml.com/ | 15:10 |
| JayF | Will you have time to address this issue rpittau or is there another volunteer who might take this action? | 15:10 |
| rpittau | converting directly to grpahviz is a pain | 15:10 |
| rpittau | JayF: not before PTG | 15:10 |
| rpittau | but we're ok until next cycle anyway | 15:11 |
| JayF | I would prefer Ironic not be the long pole in any of those requirements updates generally; but that timeline is OK in any event. | 15:11 |
| JayF | rpittau: how about 11/6 for a check-in date on that? | 15:12 |
| JayF | just to make sure we don't drop it? | 15:12 |
| rpittau | sounds good | 15:12 |
| JayF | #action rpittau to tackle bug 2026345; remove our dep on blockdiag - check in 11/6 | 15:12 |
| JayF | I'm going to basically just carry these over during action item windows each meeting | 15:12 |
| JayF | I'm doing a similar pattern in TC, it's nice to have >1 week timelines on things but still have them checked | 15:12 |
| JayF | at least for how my brain works :D if it's not useful for you please bear with me | 15:13 |
| JayF | moving on | 15:13 |
| JayF | #topic PTG Planning for PTG October 23-27, 2023 | 15:13 |
| JayF | I'm going to be taking the topics from | 15:13 |
| JayF | #link https://etherpad.opendev.org/p/ironic-ptg-october-2023 | 15:13 |
| JayF | and assigning them specific times during the windows already reserved | 15:13 |
| JayF | if you have input or requested time windows for items, or work items not listed yet, go, edit, now :D | 15:13 |
| JayF | I'm also happy to spin up a video chat and go over the scheduling sync if anyone wants to toil with me :D | 15:14 |
| JayF | anything else on PTG? | 15:14 |
| JayF | moving on | 15:15 |
| JayF | #topic Review Ironic CI status & update whiteboard if needed | 15:15 |
| JayF | I've honestly not looked at a lot of Ironic changes in the last week, I've been pretty busy -- is there anything notable going on gate-wise? | 15:16 |
| TheJulia | I spent a while on backports/fixes on friday | 15:16 |
| TheJulia | Things looked like they were in a better shape by Friday afternoon, so hopefully things are happy today | 15:16 |
| * TheJulia hopes | 15:17 | |
| JayF | nice; thank you for that | 15:17 |
| JayF | moving on? | 15:17 |
| JayF | no RFEs to review; skipping item | 15:17 |
| JayF | #topic Open Discussion | 15:17 |
| JayF | Floor is open | 15:17 |
| TheJulia | I'd <3 getting eyes on the patch to switch a job or two over to using OVN for DHCPv4 | 15:18 |
| JayF | you wanna #link it? | 15:18 |
| JayF | I think I have an old review on it, I can renew that review | 15:18 |
| TheJulia | #link https://review.opendev.org/c/openstack/ironic/+/885087 | 15:18 |
| JayF | we should get that job started sooner rather than later | 15:18 |
| iurygregory | I will add to my list | 15:19 |
| TheJulia | Once v6 support lands in neutron, I'm suspect harald will want to switch the v6 job over | 15:19 |
| TheJulia | I think he has a patch, but we're the v6 change in neutron is still in review | 15:19 |
| JayF | I think how OVS vs OVN interacts with Ironic would be a good topic | 15:20 |
| JayF | for a blog or tech talk or brown bag or whatever cliche "way to learn things" you wanna pick lol | 15:20 |
| TheJulia | for us, it is not much different, and I'm afraid it could also turn into "Ugh OVN" | 15:21 |
| TheJulia | but not bad to spread awareness | 15:21 |
| JayF | hearing 'for us it's not that different' while reading a big patch about the differences (at least in CI) is strange to hear :D | 15:22 |
| TheJulia | it is devstack setup for the differences for our virtual testing | 15:22 |
| TheJulia | because of different way OVN works. From an operator point of view, it is very similar | 15:22 |
| JayF | With my ops-y hat on (which is getting old and tattered and may not fit well anymore); I am just thinking "this is new/different services in my env, which pass data differently, which I need to know what to look for and where to look for it" | 15:23 |
| JayF | either way; just a suggestion | 15:24 |
| JayF | I +2'd the OVN patch. | 15:24 |
| JayF | We're still in open discussion; someone wanna add another thing? | 15:24 |
| TheJulia | The set power state thing with nodes is a nice issue, fwiw :) | 15:25 |
| TheJulia | with parent_nodes | 15:25 |
| TheJulia | if the parent node task is held... | 15:25 |
| JayF | I'm not sure what you're talking about without contxt? | 15:25 |
| TheJulia | eh, you can't pull a lock on a task off another task off another task in our model | 15:25 |
| TheJulia | so you can't say, if parent node has a task to run things on a child node, you can't pull a parent node task again to say "turn the power on" | 15:26 |
| TheJulia | just a fun practicality | 15:26 |
| JayF | ah, yeah | 15:26 |
| JayF | locking is hard. | 15:27 |
| TheJulia | Yeah, and there is a practicality we likely need to document | 15:27 |
| JayF | "Don't let two things touch this unless the second thing is me in a mustache" | 15:27 |
| * TheJulia makes a mental todo to do that | 15:27 | |
| JayF | Think I'm going to close up the meeting? We seem done? | 15:27 |
| TheJulia | seems so, you want people willing to discuss the ptg etherpad? | 15:28 |
| JayF | #endmeeting | 15:28 |
| opendevmeet | Meeting ended Mon Oct 9 15:28:21 2023 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:28 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2023/ironic.2023-10-09-15.02.html | 15:28 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2023/ironic.2023-10-09-15.02.txt | 15:28 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2023/ironic.2023-10-09-15.02.log.html | 15:28 |
| JayF | If someone wants to join, I can do it in a zoom | 15:28 |
| TheJulia | sure | 15:28 |
| JayF | otherwise I can just start categorizing and assigning times | 15:28 |
| JayF | aight, lemme get a thing going | 15:28 |
| JayF | https://us06web.zoom.us/j/84001967325?pwd=Ony0iqBR7jDHIpStubpAg3S0IfVIeq.1 | 15:29 |
| * JayF will brb getting headset | 15:30 | |
| dtantsur | I'll have to comment later on - no brainpower already | 15:38 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: Refactor publishing images into a new module https://review.opendev.org/c/openstack/ironic/+/897675 | 15:43 |
| dtantsur | rpittau, ^^^ I'm not yet sure if it solves us some pain with the non-bootable ISO, but it might | 15:43 |
| dtantsur | at least maybe we can avoid tying it into the existing ISO publishing logic | 15:43 |
| rpittau | ok, I'll check it ASAP | 15:45 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: [WIP] Generic API for attaching/detaching virtual media https://review.opendev.org/c/openstack/ironic/+/894918 | 15:46 |
| TheJulia | muahahahaha we're adding more topics | 15:59 |
| TheJulia | DOOOOM because it is just JayF and I :) | 15:59 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: [WIP] Extract generic image publishing code from image_utils https://review.opendev.org/c/openstack/ironic/+/897681 | 16:11 |
| dtantsur | rpittau, this is taking shape now ^^ | 16:12 |
| dtantsur | I think I'll need to finish tomorrow | 16:12 |
| rpittau | yep, I"m going offline now anyway | 16:14 |
| rpittau | see ya tomorrow! o/ | 16:14 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: [WIP] Generic API for attaching/detaching virtual media https://review.opendev.org/c/openstack/ironic/+/894918 | 16:16 |
| JayF | Detailed schedule proposed here: https://etherpad.opendev.org/p/ironic-ptg-october-2023 | 16:32 |
| -opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline momentarily while we restart it for a combined runtime and platform upgrade | 16:33 | |
| opendevreview | Derek Higgins proposed openstack/sushy-tools master: Add new ironic driver https://review.opendev.org/c/openstack/sushy-tools/+/887046 | 16:56 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Assume the parent node needs to be powered on https://review.opendev.org/c/openstack/ironic/+/896570 | 19:03 |
| stevebaker[m] | good morning | 19:14 |
| TheJulia | good morning! | 19:15 |
| TheJulia | would anyone *scream* if we added the ability for tempest to hydrate together fully functional nodes for testing | 20:05 |
| TheJulia | so we can run scenario jobs | 20:05 |
| JayF | Can you be more explicit about what you mean | 20:06 |
| JayF | you want tempest to drive the vm setup step? | 20:06 |
| TheJulia | today, tempest's model is everything is a ephermeral resource you can just spawn things | 20:06 |
| TheJulia | but we can only test ironic with pre-populated ironic nodes | 20:07 |
| TheJulia | Not drive anything locally beyond "posting to /v1/nodes and creating the nodes it knows about" | 20:07 |
| JayF | so you're saying, ensure the BM nodes exist | 20:07 |
| JayF | then tempest does the creates? | 20:08 |
| JayF | or at least, *can* perform creates if we want? | 20:08 |
| TheJulia | yes, *can* | 20:08 |
| JayF | I have an alternate suggestion I was going to bring up during the PTG session about this | 20:08 |
| JayF | what would that test that we couldn't use a fake-driver-node to test? | 20:08 |
| TheJulia | we do that for what we can already | 20:08 |
| JayF | (including the possibility of enhancing the fake drivers) | 20:08 |
| TheJulia | it is scenario integration tests where we cannot | 20:08 |
| TheJulia | since we're driving the full flow | 20:09 |
| JayF | I don't know the specific meaning of > scenario integration tests | 20:09 |
| TheJulia | ironic_tempest_plugin.tests.scenario.* | 20:10 |
| TheJulia | where we drive deployments and whatnot to ensure expected integraiton/behavior worsk | 20:10 |
| JayF | Why does having tempest do the creates there matter? Like what is the heart of the reason you wanna change the behavior? | 20:10 |
| TheJulia | folks don't want an intermediate step to do anything perceived as customization | 20:11 |
| JayF | I mean, I can see how it would make the tests cleaner, although I wonder who these mysterious folks are :P | 20:12 |
| TheJulia | basically folks downstream just want to be able to drop a tempest config in place and not have to do baremetal node create commands | 20:12 |
| TheJulia | basically | 20:12 |
| JayF | having the tests rely on weird existing system state also seems not awesome too; it'd be nice if we could somehow have devstack tell tempest what exists there | 20:12 |
| JayF | rather than passing around assumptions | 20:12 |
| TheJulia | problem is we're aslo trying to deal with the existing delination model in tempest which has made things difficult | 20:13 |
| TheJulia | you can create a VM, you can't create a baremetal machine out of thin-ish air | 20:14 |
| JayF | Like my gut feel on this is that if we want tempest to create the VMs, we should ensure there's some kinda inventory or information passed down that it reads to create them from (at least in the devstack case?) | 20:14 |
| JayF | let me restate | 20:15 |
| JayF | what can we do to ensure that tempest test node creates and the reality of the nodes that devstack created stays aligned? | 20:15 |
| TheJulia | creating VMs locally would violate the "it can't touch the substrate rules" | 20:18 |
| JayF | I'm saying, if tempest is creating the test nodes | 20:18 |
| JayF | node creates being API node creates | 20:19 |
| JayF | reality of nodes devstack created == fake bare metal just waiting for tempest to create their API friends | 20:19 |
| JayF | re-restated: what can we do to ensure that tempest test /v1/node creates and the reality of the libvirt vm+[sushy-tools/vbmc/vpdu] machines that devstack created stays aligned? | 20:19 |
| JayF | (and, for that second half, maybe for your downstream "the reality of the physical machines you hook up to the other side of tempest" is part of that question?) | 20:20 |
| TheJulia | I'm thinking entirely disjoint | 20:21 |
| TheJulia | like "hey, tempest, here are your loads to perform, pre-staged" | 20:21 |
| JayF | I am asking 100% about ^^^ how that message is passed | 20:21 |
| JayF | and making sure it's not "we just read the same env vars and make the same assumptions" | 20:21 |
| JayF | and instead is more like an inventory file of sorts | 20:22 |
| TheJulia | To me, that is an implementation detail, but I was thinking a json or yaml payload of some sort | 20:22 |
| JayF | What that payload looks like is an implementation detail; that is exists and is specified in a flexible way is what I care about :D | 20:22 |
| TheJulia | most likely $details we would just post to the API to create a node | 20:23 |
| TheJulia | and any ports | 20:23 |
| TheJulia | for that node | 20:23 |
| JayF | that seems to punt any testing benefit we'd get from it but still gets the mechanism improvement for your downstream | 20:24 |
| JayF | seems okay but if we ever wanted to push that payload forward it'd be in devstack *and* i-t-p instead of one or the other, I guess | 20:24 |
| JayF | but that's probably what'd happen anyway? | 20:24 |
| TheJulia | upstream wise, we could just defer the creation, or not. dunno, it is just a slightly complex bit of the flow for upstream CI | 20:24 |
| JayF | oh, I'm fine with that as a model | 20:24 |
| JayF | I was just trying to solve use cases that may not exist | 20:25 |
| JayF | e.g. "you provide something that looks inventory-like to tempest and it tests against it" being a solution for testing against real-ish hardware | 20:25 |
| TheJulia | I guess the challenge is we have a couple tests where we have many nodes, and most have at least 2 for reasons | 20:25 |
| JayF | but that's not an actual problem to solve | 20:25 |
| JayF | it's just a made up one :D | 20:25 |
| TheJulia | and completely removing is sort of a PITA efficiency wise... hmmm | 20:25 |
| JayF | like if tempest is integration tests | 20:26 |
| TheJulia | "create_nodes_if_not_exist"! | 20:26 |
| JayF | that inventory file looking like what some customers would get from a hardware vendor just feels nice | 20:26 |
| JayF | but how to structure this kinda code is the tech that I'm probably the worst at | 20:26 |
| TheJulia | yeah, we do similar with bifrost | 20:26 |
| JayF | so just take all this as ... ideas to see if any of them stick for you :D | 20:26 |
| TheJulia | "take this, do a thing!" | 20:26 |
| JayF | that's mentally what I was modelling | 20:27 |
| JayF | but the bifrost inventory leaks a lot of Ironic API details :D | 20:27 |
| TheJulia | It was never meant to pave over everything to be an independent clean surface | 20:27 |
| JayF | oh, I just meant for terms of referencing this for this | 20:28 |
| opendevreview | Verification of a change to openstack/ironic-python-agent master failed: Add mlnx deploy_step entry to enable deploy time firmware https://review.opendev.org/c/openstack/ironic-python-agent/+/893371 | 22:36 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!