| opendevreview | Steve Baker proposed openstack/networking-generic-switch master: Create service plugin to configure security groups https://review.opendev.org/c/openstack/networking-generic-switch/+/949834 | 03:22 |
|---|---|---|
| opendevreview | Steve Baker proposed openstack/networking-generic-switch master: Support security groups in netmiko base class https://review.opendev.org/c/openstack/networking-generic-switch/+/949835 | 03:22 |
| opendevreview | Syed Haseeb Ahmed proposed openstack/ironic bugfix/27.0: Control port updates with update_pxe_enabled flag https://review.opendev.org/c/openstack/ironic/+/951719 | 04:37 |
| opendevreview | Michal Nasiadka proposed openstack/networking-generic-switch master: doc: Rework support matrix for trunk driver https://review.opendev.org/c/openstack/networking-generic-switch/+/942338 | 05:23 |
| opendevreview | Michal Nasiadka proposed openstack/networking-generic-switch master: doc: Rework support matrix for trunk driver https://review.opendev.org/c/openstack/networking-generic-switch/+/942338 | 05:24 |
| dtantsur | Looking at the changelog, I've just realized that we're reaching API version 1.100 next :) | 10:43 |
| iurygregory | yay \o/ | 11:07 |
| opendevreview | Queensly Kyerewaa Acheampongmaa proposed openstack/sushy master: Implement set_datetime method in Manager and refresh local state after PATCH https://review.opendev.org/c/openstack/sushy/+/950539 | 11:10 |
| iurygregory | I like that the definition for FirmwareInventory in UpdateService is the same of SoftwareInventory | 11:22 |
| dtantsur | well, firmware is software in a sense :) | 11:22 |
| iurygregory | yup =) | 11:24 |
| opendevreview | Queensly Kyerewaa Acheampongmaa proposed openstack/sushy-tools master: Add PATCH support for Redfish DateTime fields in Manager resource https://review.opendev.org/c/openstack/sushy-tools/+/950925 | 12:37 |
| opendevreview | Abhishek Bongale proposed openstack/ironic-tempest-plugin master: Add Tempest tests for inspection rules in Ironic https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/951761 | 12:38 |
| opendevreview | Julia Kreger proposed openstack/ironic master: trivial: revise text from mutlinode job changes https://review.opendev.org/c/openstack/ironic/+/951766 | 13:11 |
| JayF | dtantsur: fwiw I already checked, there are other openstack apis with 1.100 so we aren't groundbreaking :) | 13:32 |
| dtantsur | damn, I hoped we would be the first :) | 13:32 |
| JayF | you are less afraid of old code than I am ;) | 13:34 |
| cardoe | Just wanna point folks at https://review.opendev.org/q/hashtag:%22ironic-week-prio%22+AND+status:open | 14:03 |
| iurygregory | dtantsur, I've found a funny issue when requesting information about a resource in the FirmwareInventory , my code was returning sushy.exceptions.ResourceNotFoundError: Resource https://<BMC>/redfish/v1/UpdateService/FirmwareInventory/21 not found, after doing some request in a row for the endpoint https://paste.opendev.org/show/bXtMYD8BqOr4Q1rSY6lg/ <sweet>! | 14:21 |
| TheJulia | cid: w/r/t https://review.opendev.org/c/openstack/ironic/+/949681 if you want to post a quick follow-up, I'd be good with changing my vote | 14:25 |
| cid | Sure, update in a few. | 14:32 |
| opendevreview | cid proposed openstack/ironic master: Add a new configuration option, ``bootloader_by_arch`` https://review.opendev.org/c/openstack/ironic/+/949681 | 14:37 |
| * cid wonders why I left that hanging for that long. | 14:38 | |
| cid | Thanks, TheJulia | 14:38 |
| TheJulia | no worries! | 14:38 |
| dtantsur | iurygregory: this is impressively terrible | 14:52 |
| iurygregory | dtantsur, 100% | 14:52 |
| iurygregory | going to test on the dell to see if happens :D | 14:52 |
| opendevreview | Merged openstack/ironic master: ci: Expand the multinode job -- remove tinyipa usage https://review.opendev.org/c/openstack/ironic/+/950206 | 14:55 |
| TheJulia | wooohoooo! | 15:09 |
| iurygregory | at least doesnt seem to happen on Dell, only in the HPE I have (sweet) | 15:10 |
| TheJulia | cardoe: fyi, https://review.opendev.org/c/openstack/ironic/+/946677 :) | 15:15 |
| cardoe | I'll review that today. | 15:19 |
| TheJulia | Thanks! | 15:21 |
| opendevreview | Verification of a change to openstack/bifrost master failed: Fix python interpreter when installing in venv https://review.opendev.org/c/openstack/bifrost/+/949518 | 15:40 |
| opendevreview | Merged openstack/networking-generic-switch master: Make devstack runs slightly more deterministic https://review.opendev.org/c/openstack/networking-generic-switch/+/951027 | 15:58 |
| opendevreview | Merged openstack/ironic-python-agent master: Remove unused tinyipa CI job https://review.opendev.org/c/openstack/ironic-python-agent/+/951595 | 16:08 |
| opendevreview | Verification of a change to openstack/ironic stable/2025.1 failed: Fix agent get_XXX_steps retries from being treated as not fresh agents https://review.opendev.org/c/openstack/ironic/+/950308 | 16:14 |
| iurygregory | dtantsur,do you think it would be ok of having sushy providing the list of names of the resources that have updatable == true in the SoftwareInventoryCollection https://github.com/openstack/sushy/blob/master/sushy/resources/updateservice/softwareinventory.py#L78 ? | 16:27 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Automated cleaning by runbook https://review.opendev.org/c/openstack/ironic/+/945259 | 17:11 |
| opendevreview | Merged openstack/ironic master: api: Add schema for node firmware API https://review.opendev.org/c/openstack/ironic/+/945943 | 17:30 |
| opendevreview | Merged openstack/ironic stable/2025.1: Allow to unprovision instance from service wait states https://review.opendev.org/c/openstack/ironic/+/951682 | 17:34 |
| opendevreview | Merged openstack/ironic stable/2025.1: Control port updates with update_pxe_enabled flag https://review.opendev.org/c/openstack/ironic/+/951631 | 17:35 |
| opendevreview | Verification of a change to openstack/ironic master failed: Make port binding failure (configurably) fatal https://review.opendev.org/c/openstack/ironic/+/699953 | 17:54 |
| opendevreview | Merged openstack/networking-generic-switch master: Improve Netmiko Device Commands Documentation https://review.opendev.org/c/openstack/networking-generic-switch/+/951659 | 17:55 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: [WIP] Eventlet: Migrate API & JSON-RPC to cheroot https://review.opendev.org/c/openstack/ironic/+/951054 | 18:23 |
| JayF | fwiw, c i d and I managed to track down issues with ^ relating to single-process and json-rpc mode; single-process is fixed and there's a lead for json-rpc | 18:42 |
| JayF | jobs using rabbit just ... passed | 18:42 |
| opendevreview | Verification of a change to openstack/ironic master failed: Make port binding failure (configurably) fatal https://review.opendev.org/c/openstack/ironic/+/699953 | 19:03 |
| opendevreview | Merged openstack/networking-generic-switch master: Cast numeric Netmiko kwargs to native types. https://review.opendev.org/c/openstack/networking-generic-switch/+/951724 | 19:39 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Automated cleaning by runbook https://review.opendev.org/c/openstack/ironic/+/945259 | 19:46 |
| * JayF does the it's done dance | 19:46 | |
| cardoe | runbooks officially went out with 2025.1 right? | 20:22 |
| cardoe | I guess we should use major versions and not OS releases. | 20:24 |
| opendevreview | Doug Goldstein proposed openstack/ironic-specs master: move a number of specs to their respective Ironic release https://review.opendev.org/c/openstack/ironic-specs/+/951812 | 20:37 |
| cardoe | boy that side bar... https://specs.openstack.org/openstack/ironic-specs/index.html stinks its missing the versions | 20:38 |
| TheJulia | dtantsur: any response regarding systemd w/r/t your comments on https://review.opendev.org/c/openstack/oslo.service/+/945720/31/oslo_service/backend/threading/service.py ? | 21:00 |
| TheJulia | JayF: cool cool | 21:00 |
| TheJulia | dtantsur: asking because I'm working on a list of downstream stories to task/assign out regarding eventlet stuffs and wants to sync up there. | 21:01 |
| JayF | TheJulia: I would not be surprised if we have cid's wsgi-removal thing passing tests (at least superficially; I know there's still feedback on that patch w/r/t configs) in a handful of days. I think we ID'd the root cause of the frreze and all. | 21:02 |
| JayF | I am just shocked at how much progress we're making | 21:02 |
| TheJulia | we've still got a ways to go | 21:03 |
| TheJulia | like found an item not on Dmitry's list, but it has to do with the vnc proxy stuffs | 21:04 |
| JayF | I know the liklihood of a hard problem goes up the closer we get to removal | 21:04 |
| JayF | but it feels real/achievable to get rid of it for the first time | 21:04 |
| TheJulia | Yeah, I think the ssl configs are more so a consensus blocker since the exsting parameter/keys are misplaced and marked for removal | 21:09 |
| TheJulia | But yeah, it seems like logical chunks are left. Some of it is intertwined though | 21:10 |
| JayF | honestly I mainly care more about getting the technical bits working | 21:10 |
| JayF | moving config around is something I know I can do :) | 21:10 |
| TheJulia | fair | 21:10 |
| TheJulia | Anyway, I'm going to call it a day shortly, my brain is reaching molten brain | 21:13 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Automated cleaning by runbook https://review.opendev.org/c/openstack/ironic/+/945259 | 21:24 |
| opendevreview | Merged openstack/ironic master: Make port binding failure (configurably) fatal https://review.opendev.org/c/openstack/ironic/+/699953 | 21:39 |
| cardoe | 34 changes with ironic-week-prio. Some of which are old and have merge conflicts. Aiming to get the good stuff in and get that list down under 25 this week. | 22:21 |
| JayF | some of those are stable patches waiting for their counterparts to merge in fresher branches | 22:23 |
| cardoe | Yep. That’s what I meant by the good stuff. | 22:27 |
| JayF | I'll note, we've also merged 14 patches into ironic-proper in the last week, which is killer | 22:29 |
| JayF | upstream productivity is up! :D | 22:29 |
| opendevreview | Verification of a change to openstack/ironic stable/2025.1 failed: Fix agent get_XXX_steps retries from being treated as not fresh agents https://review.opendev.org/c/openstack/ironic/+/950308 | 22:31 |
| cardoe | The flogging will continue until the queue is empty! | 22:34 |
| cardoe | TheJulia: so the config drive stuff. I’m on mobile so hard to comment. But it’s okay. But you’re changing the IDs which is actually problematic. This is something I’m trying to hash out with nova right now as well. Because we have code in there to fix up the config drive. So it would make sense if the two behaved the same. | 22:36 |
| TheJulia | cardoe: how is it problematic? | 22:37 |
| cardoe | But physical ports and bonds are done wrong. And cloud-init has for ages worked around OpenStack. | 22:37 |
| cardoe | So there’s a schema and we don’t follow it. | 22:37 |
| TheJulia | I'm not entirely sure about that, but specific pointers would help since we generate in the same style from what I've seen at a glance | 22:37 |
| cardoe | So they literally load our network_data.json fix it up and then use it. | 22:38 |
| cardoe | I’ll share in the morning if that’s okay. | 22:38 |
| TheJulia | cloud-init, eh? thats easy to fix | 22:38 |
| cardoe | But I’ve raised a bug with cloud-init and there’s been finger pointing. | 22:39 |
| TheJulia | but the IDs we can't reconcile to match it exactly because on the nova side is is like tap-something | 22:39 |
| cardoe | Yeah nova is wrong. | 22:39 |
| TheJulia | and we get none of that from nova unless we just regenerate it | 22:39 |
| TheJulia | well, regenerate it because the mtu is null | 22:39 |
| TheJulia | which sucks that is a possible case | 22:39 |
| TheJulia | (like, seriously, how!) | 22:39 |
| cardoe | So like when you use vlan tags is where it blows up. | 22:39 |
| TheJulia | oh yeah, for sure | 22:39 |
| cardoe | Cause it appends the number to the id and it must be less than 15 characters. | 22:40 |
| TheJulia | yeah | 22:40 |
| TheJulia | I got bitten by that elsewhere recently as well | 22:40 |
| cardoe | ID can be anything ya want if you have a name field. | 22:40 |
| TheJulia | correct | 22:40 |
| TheJulia | the reader doesn't have to use the name we give it, it just has to assemble the thing from my point of view | 22:40 |
| cardoe | And there’s an option to say use kernel naming and append the vlan number. | 22:40 |
| TheJulia | yeah, that we also have no idea what the workload is | 22:41 |
| cardoe | Which is what I think we should use. | 22:41 |
| cardoe | So that we can then use any value we want for ID | 22:41 |
| TheJulia | I'm not sure that is really possible, but send me the links and a clear comment and I can dig into it tomorrow | 22:42 |
| cardoe | So happy to +2 it but I think we should revisit and sync the behavior with the ironic driver in nova. | 22:42 |
| TheJulia | nova is basically making clear they don't care about this area and I'd rather just fix up bad data when we can spot it, tbh | 22:42 |
| TheJulia | the nova behavior is also just wrong too, as noted | 22:43 |
| cardoe | https://github.com/canonical/cloud-init/blob/main/cloudinit/config/schemas/schema-network-config-v1.json | 22:43 |
| cardoe | null is valid for MTU btw. | 22:43 |
| TheJulia | so for cloud-init specifically | 22:46 |
| TheJulia | if we hand it it's preferred payload | 22:46 |
| TheJulia | (and still bonkers) | 22:47 |
| TheJulia | :) | 22:47 |
| cardoe | https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/helpers/openstack.py line 530 and down is where their fixups are. | 22:48 |
| cardoe | https://github.com/canonical/cloud-init/issues/6018 | 22:49 |
| TheJulia | yeah, different platform specific schema to cloud-init specific yaml | 22:49 |
| cardoe | Yep. But they don’t use the regular loader to their yaml cause sometimes nova doesn’t conform to the schema. | 22:50 |
| cardoe | I linked the wrong schema. That was their schema. | 22:50 |
| cardoe | We have schemas for config drive. | 22:51 |
| TheJulia | Anyhow, I'm not opposed to improving it or even eventually doing a separate cloud-init yaml conversion, but I think the shortish term key is consistency within openstack because cloud-init is no the only tool consuming it, even though cloud-init is the big kid on the block | 22:51 |
| cardoe | Yep. I’m a +2. But I think we need to have the nova ironic behavior the same. And there’s some improvements. | 22:51 |
| TheJulia | lets detail the specific issues as bugs to track and address then, because i know there are differences, and definitely cases we should improve/guard against | 22:53 |
| TheJulia | and doing that in the first patch is better but we might want to have patches for some of those behaviors before determining we have a missing MTU on the payload | 22:53 |
| cardoe | Vasyl’s made those patches like https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/940678 | 22:53 |
| TheJulia | (and that *has* bitten folks trying to pass a null mtu along with neutron because neutron can have constrained mtus | 22:54 |
| cardoe | And is getting bit | 22:54 |
| cardoe | https://docs.openstack.org/nova/latest/_downloads/9119ca7ac90aa2990e762c08baea3a36/network_data.json | 22:55 |
| cardoe | That’s the schema I was looking for. | 22:55 |
| cardoe | So you’ll see they have l2_id which their exa | 22:57 |
| cardoe | Example is eth0 | 22:57 |
| cardoe | They generate the tap value because it needs to conform to kernel naming. | 22:59 |
| cardoe | But our ironic nova doesn’t follow that. | 23:01 |
| TheJulia | okay, yeah, I see what you mean | 23:01 |
| TheJulia | eww | 23:01 |
| TheJulia | hmmmmm | 23:02 |
| cardoe | Then the other value is supposed to be a UUID. The ironic nova code does that. But the regular nova code doesn’t. | 23:02 |
| TheJulia | so, I've never seen the id value ever be used as an assumption matching to interface name | 23:03 |
| cardoe | The cloud-unit people told me to tell nova to conform to its own published schema | 23:03 |
| TheJulia | The expectation, when I first worked iwth the config drives like a decade ago... was it was just for cross-referencing | 23:03 |
| TheJulia | fair enough | 23:03 |
| cardoe | The os-vif repo has code that does. | 23:03 |
| TheJulia | so, ultimately the issue is cloud-init wants to name interfaces using the id field ? | 23:04 |
| cardoe | Which is used by every hypervisor backend except ironic | 23:04 |
| cardoe | No. They want to name it by the name field. | 23:05 |
| cardoe | But nova doesn’t publish a name field | 23:05 |
| cardoe | Which is required per the schema. | 23:06 |
| TheJulia | So that is the issue I think is super problmatic since we can't know what the workload is or how it is operating | 23:06 |
| cardoe | Right. | 23:08 |
| cardoe | I suspect that’s why name went away. | 23:08 |
| TheJulia | yeah | 23:08 |
| cardoe | I think a user controlled name isn’t bad. | 23:09 |
| TheJulia | and strictly speaking I don't think a name is required by that field, but the example suggests id is a name so its super confusing | 23:09 |
| cardoe | Attach an interface to network X and call it mgmt0 would be cool. | 23:09 |
| TheJulia | except, vlan_link which is also a name | 23:10 |
| cardoe | Yep | 23:10 |
| TheJulia | WHY! | 23:10 |
| * TheJulia wonders if pugs were involved with the design of that schema | 23:11 | |
| * TheJulia knows the answer already | 23:11 | |
| * TheJulia notes references to pugs is going back waaaaay back in time | 23:11 | |
| cardoe | Anyway. Vasyl and I want ironic to return data that always works. | 23:12 |
| cardoe | To comply with canonicals licensing, we have to do a vendor data script to jump through some hoops | 23:12 |
| cardoe | vseinko I think is his IRC. I don’t know who he works for. But we’ve both hit the same issue. | 23:13 |
| TheJulia | I'm not sure we can realistically do that in cases we're fixing up much less unless someone just tells us what naming format to use and even then they would need to know the internals. I guess this is why I like glean, it doesn't care about id values except to match links and networks together | 23:14 |
| cardoe | Like I said +2 on your change cause it makes it better. And makes the code match nova for the config drive generation. | 23:14 |
| TheJulia | Yeah, I've known vasyl for a long time | 23:14 |
| TheJulia | Anyhow, we should add some bugs just so we know its an issue and.. can at a minimum record/identify challenges and maybe figure out a cleanish solution at some point. | 23:19 |
| TheJulia | And with that, I'm going to run into town with the wife | 23:20 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!