| rpittau | good morning ironic! o/ | 04:10 |
|---|---|---|
| TheJulia | It is always morning ;) | 04:13 |
| TheJulia | Even when there is a game being played! https://usercontent.irccloud-cdn.com/file/dsfliZNP/1712808854.JPG | 04:14 |
| JayF | ptg session in 15, yeah? | 04:46 |
| * JayF is trying to keep awake | 04:46 | |
| TheJulia | in 10 minutes! | 04:50 |
| rpittau | iurygregory: as a rule of thumb, usually anything that is voting in check is also voting in gate | 04:54 |
| JayF | ++ | 04:55 |
| rpittau | 5 minutes to PTG session two-and-a-half! https://meetpad.opendev.org/apr2024-ptg-ironic | 04:56 |
| TheJulia | Goodnight, see y'all in the morning | 05:44 |
| rpittau | thanks TheJulia, see you later :) | 05:46 |
| rpittau | breakfast time! | 05:46 |
| dtantsur | JayF: could you add me to the ARM invite? I cannot promise to make 1700 UTC, but I will try. | 06:34 |
| dtantsur | Also, folks, could I get some eyes on https://review.opendev.org/c/openstack/ironic/+/915415 (especially people caring about servicing)? | 06:43 |
| rpittau | JayF: please add me too to the ARM invite, thanks! | 06:48 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: Fix get_async_step_return_state to account for servicing https://review.opendev.org/c/openstack/ironic/+/915469 | 08:59 |
| dtantsur | another small but blocking issue in servicing ^^ | 08:59 |
| dtantsur | oh lovely, our unit tests rely on the broken behaviour.... | 11:39 |
| iurygregory | good morning | 11:41 |
| iurygregory | "expected" I would say =( | 11:41 |
| iurygregory | I wasn't able to stay awake till 2am, I lost the PTG session =( | 11:42 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: Fix get_async_step_return_state to account for servicing https://review.opendev.org/c/openstack/ironic/+/915469 | 11:42 |
| dtantsur | okay, this does less than I'd want to, but it should fix the issue. ^^^ cc janders | 11:42 |
| iurygregory | rpittau, yeah I know, was wondering if it was wondering if it was worth having in both check and gate tbh | 11:43 |
| dtantsur | iurygregory, JayF, voting jobs MUST be in the gate. It's not optional. | 11:50 |
| dtantsur | good catch | 11:51 |
| iurygregory | in my mind we could since it was just codespell, but it's life =) | 11:51 |
| dtantsur | It's basically a rule. Otherwise, you can let a change merge that breaks the CI | 11:52 |
| iurygregory | makes sense | 11:54 |
| opendevreview | Merged openstack/ironic master: Stop assuming service steps have priorities https://review.opendev.org/c/openstack/ironic/+/915415 | 12:37 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic stable/2024.1: Stop assuming service steps have priorities https://review.opendev.org/c/openstack/ironic/+/915357 | 12:40 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic stable/2023.2: Stop assuming service steps have priorities https://review.opendev.org/c/openstack/ironic/+/915358 | 12:40 |
| rpittau | 10 minutes to PTG ! https://ptg.opendev.org/ptg.html | 12:51 |
| TheJulia | good morning | 12:55 |
| JayF | Running a little behind, but don't worry about waiting for me for secure boot | 13:04 |
| JayF | rpittau: I see the update on PTG board, are we reconvening for networking before the top of the hour? | 13:36 |
| rpittau | JayF: we're on a long pause, we finished secure boot earlier than expected | 13:37 |
| JayF | I sorta suspected that might be the case when I saw the schedule :D | 13:37 |
| JayF | I'll jump in just before top of the hr then | 13:37 |
| iurygregory | we are just waiting for you JayF \o/ | 13:38 |
| rpittau | JayF: if you want to jump in now we're just telling stories :) | 13:38 |
| samcat116 | I can't join the ptg call unfortunately (on a train) but have some networking things to add to the current topic | 14:09 |
| TheJulia | samcat116: Greetings! | 14:16 |
| samcat116 | 👋 | 14:17 |
| TheJulia | wasamadoodle <-- newest ironic word | 14:18 |
| TheJulia | samcat116: how long will you be on the train for? I suspect we could always revisit the discussion | 14:21 |
| samcat116 | A while unfortunately :( | 14:25 |
| samcat116 | But I think all thats in there is a good start | 14:25 |
| TheJulia | samcat116: I think we're going to send a couple different mailing list posts on this subject, we're collecting action items right now | 14:27 |
| TheJulia | and discussion is still ongoing | 14:37 |
| SvenKieske | TheJulia: regarding the interop PTG stuff, I just dropped some - hopefully helpful - stuff in the etherpad (I guess I can't attend). maybe we can chat about this later, you can also reach me via usual channel (linkedin, mail, matrix or here) | 15:44 |
| SvenKieske | sorry for hijacking this channel :) | 15:44 |
| JayF | you are always welcome to hijack the channel :D | 15:45 |
| SvenKieske | thanks :D | 15:45 |
| TheJulia | SvenKieske: would it make sense to try and setup a time? It looks like it is just Tobias and I right now. | 16:09 |
| SvenKieske | sure, I send you a private message | 16:13 |
| JayF | tc ptg started | 16:14 |
| dtantsur | We're done with ironic | 16:14 |
| rpittau | ironic PTG has ended, again thank you everyone! | 16:14 |
| iurygregory | \o/ | 16:14 |
| dking | I want to be sure to clarify, nothing in the removal of the Inspector is going to impact the use of custom collectors in IPA? | 16:16 |
| dking | We use it in an undocumented way. We let unknown nodes PXE boot, which defaults to inspection, and this allows us to run a custom collector, which in turn sets off a process that enrolls the node into Metal3 Baremetal Operator, which is what ultimately gets it into Ironic. It ends up running inspection twice. | 16:18 |
| masghar | Most of the collectors have been moved into ironic | 16:18 |
| dtantsur | dking: nothing. If you use hooks (server-side plugins), they'll need a bit of adjustments | 16:18 |
| dtantsur | masghar: you're talking about hooks, dking is asking about IPA-side plugins | 16:18 |
| dtantsur | dking: wwwwwwwow. Maybe we should re-open the conversation about auto-discovery in Metal3... but this is going to work as before. | 16:19 |
| dking | We make heavy use of a custom Hardware Manager. | 16:19 |
| dking | Yeah, it's probably not the standard use case. Back when I was in OpenMetal, we used the inspect to enroll features, but since that wouldn't get it into BMO, I decided to do this crazy stuff. | 16:20 |
| dtantsur | Unrelated, but I'd still appreciate a review of https://review.opendev.org/c/openstack/ironic/+/915469, it's blocking many practical applications of servicing. | 16:20 |
| masghar | dtantsur: yep, I mixed them up | 16:21 |
| dtantsur | dking: how do you enroll credentials btw? | 16:21 |
| dking | dtantsur: If there is a chance for metal3 to do auto-discovery, please let me know. | 16:21 |
| dtantsur | dking: for your context, the previous discovery discussion was https://github.com/metal3-io/baremetal-operator/issues/1124, and it was closed to a large extent because of different people pulling in different directions | 16:22 |
| dtantsur | The usual key problem is knowing credentials for newly enrolled nodes. | 16:22 |
| dking | dtantsur: We have a backend that the new nodes hit from the custom collector. The process is still being improved, but we interface with things like the CMDB. | 16:22 |
| dtantsur | dking: hmm, that's a bit tough to make generic enough to include upstream | 16:23 |
| dtantsur | It could be reshaped to this. Ironic enrolls the new node. BMO picks it up automatically and lets it sit in the "unmanageable" state. Your (new) controller watches for such hosts and updates them from a CMDB. | 16:24 |
| dking | Yeah, there's not a great solution for the creds. It can get pretty hacky. The nodes themselves generate the credentials, but I don't know of a way to get those creds into Ironic directly if it hasn't been enrolled yet. In the current process, they get piped out and eventually end up being sent to BMO. | 16:24 |
| dtantsur | If you come from such an angle, it seems a feature that we could accept. | 16:25 |
| dtantsur | Depends also on whether you're up for some Go coding | 16:26 |
| dking | I'm still a Go newbie, but I've been looking for a project to write in it. | 16:27 |
| dking | I don't really know how much of our current use case is good for generic users, though. It's currently tightly coupled with CMDB, so it would have to take some thought. I might need to look through the above issue to see what people were proposing. | 16:28 |
| dtantsur | Yeah, see my idea above for how to decouple them. Needs more thoughts of course. | 16:30 |
| dking | dtantsur: Just spitballing here, but could a BMO custom collector be useful? You can't get the creds from Ironic directly, but if you set the creds on the server, you could push to some some location that is accessible to BMO. Perhaps it would involve Vault somehow? | 16:35 |
| dtantsur | dking: I've just written https://github.com/metal3-io/baremetal-operator/issues/1124#issuecomment-2050092410 to leave a track of our conversation | 16:39 |
| dtantsur | The core of my idea is to leave the credentials completely to the downstream code. It can be your controller that takes them from the CMDB or some other place. | 16:40 |
| dtantsur | I really need to disconnect now, happy to continue here or on the bug tomorrow. | 16:41 |
| dking | Have a good night! | 16:43 |
| cid | good night \o | 19:56 |
| cid | By the way, I enjoyed (maybe not the right word), the PTG session today. I hope to join more in the future. | 19:57 |
| JayF | o/ thanks, you're always welcome | 20:00 |
| JayF | Note that we need to look at rootwrap->privsec migration, I think? Or update docs to remove refernce to rootwrap | 21:39 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
