Meeting started at 2025-12-03 16:05:04 UTC

The Meeting name is 'CentOS Hyperscale SIG'

morning folks

!hi

Michel Lind (salimma) - he / him / his

!hi

Raymond Colebaugh (rcolebaugh) - he / him / his

Good morning!

!hi

Neal Gompa (ngompa) - he / him / his

let's get started

any followups from the last meeting?

I think we brought up perf and ynl? In that case I have updates

!hi

Naraian K (naraiank) - he / him / his

welcome K Naraian (He / Him / His) ('The Exorcist'), what's your interest in Hyperscale?

and we have something for announcement / open floor related to this too, not sure if that was discussed in a meeting before

!hi

Jonathan Wright (jonathanspw)

I'm basically a DB guy. I'm just following the event.

I'm in Testing and Quality SIG, So, I'm open to contribute there..

we could definitely use more help with testing & quality SIG - what we'd like, given Hyperscale has signed testing packages, is to have a flow similar to Bodhi where people can test drive updates and give feedback before getting them promoted

hope you find this interesting

anyone else has follow-ups? and hi again Jonathan

It's been a while!

Schedules finally meshed for me and this meeting.

I think we can move to

Michel Lind UTC: you want to announce something?

yes

trying to get some traction with some of the refactoring we're doing, and hopefully get some EL ecosystem changes in getting it easier to rebase both the kernel and userspace tooling so we can override specific bits in SIGs

from our experience, having all the userspace tooling bundled made Neal's job hell rebasing (because they might not compile on older EL) and makes my job hard iterating to add features and dealing with installability issues

yeah

it's been a while since the two of us have done a talk together, so I'm excited :)

Summit, wasn't it? wait no

F43 release party

"a while" :P

oh not dec 15?

Michel Lind UTC: in person together :)

technically december 15 11:59

but that's midnight dec 16

so afternoon for US folks

yup

let's try not to submit on the last day so it's irrelevant :)

indeed

Anyone else have anything for announcements?

doesn't look like it

we still need to migrate off pagure at some point

have we figured out what the migration looks like yet?

run the tool and call it a day?

I suppose that works :D

the only unpleasant part is the lack of username bindings across pagure to gitlab

so the posts aren't associated with you anymore

but meh

Davide Cavalca: you want to just do it?

we already cleaned up the backlog a while back

yeah, probably not today though

sure

we need to purge the member rolls

there are so many inactive users I can't use the bot commands to poke everyone for meetings

yeah... I have been a bit remiss in scripting an activity check tool

if anyone wants to take a stab, so you're not blocked on me, feel free to take point

but hopefully things will be quiet when a lot of people are going to LPC :)

anything else for this week?

I've got one thing to bubble up

I'm not sure if Jonathan Wright was going to bring this up or not, but this came to ALESCo's desk a while back: https://github.com/AlmaLinux/ALESCo/pull/15

submitted by james

I wasn't, but I guess it's relevant, honestly.

The front-runner for what to use as a base is the hyperscale kernel.

this seems reasonable

most likely this will involve maintaining a small packaging delta so that it's called `kernel-hs` or `kernel-hyperscale` in AlmaLinux

most likely this will involve them maintaining a small packaging delta so that it's called `kernel-hs` or `kernel-hyperscale` in AlmaLinux

which circles back a bit to Michel Lind UTC and I talking about kernel packaging improvements

will this be in a separate repo or the same repo?

will be funny if that is the solution to Hyperscale users wanting a signed kernel - just get it from another source

I don't know, given that it will be renamed, it can exist either in the main repo or a separate one

Conan Kudowe still need to do that small meeting with Alma to hammer out details prior to Connect, ideally, so we don't waste time when we meet in person

it probably will be a separate one, tbh

esp since there's this RC one

*RFC

yes

includepkgs is an option too anyway :) so either way it does not matter

that was why I wanted to bring it up

because we're going to need to prioritize getting that worked out

since none of us involved are going to LPC iirc it makes meeting easier hopefully

yup

I can try coming to alesco too and drop my conflicting meeting, but not tomorrow - can't afford to miss that one

though I don't promise to not staycation somewhere and wallow in the misery of yet again missing a chance to go to Japan

oh I plan to staycation too

but probably closer to Christmas -- well not too close in case rates start shooting up

yeah

if nobody else has anything to bring up we probably should not hog the room though :)

we should follow up on this topic elsewhere

one last thing to bring up before we go

probably separate, maybe synergy, maybe its own repo.

Fedora merged my pull request to add the _ability_ to set build counters for package builds

so now the door is open to leveraging this functionality in other places... we might want to deliberately backport this into Hyperscale

I'm working on a simple PoC script that demonstrates how the interface should be used

oh very nice

then we can bug build systems on supporting it

yeah, would probably be easier to demonstrate it in use in "production" in HS

assuming it does not need Koji code changes and only config

it needs koji code work :)

every build system will need code work to leverage it

I filed a ticket with the Open Build Service folks for them to support it a couple weeks ago

I intend to make similar proposals to ALBS, Koji, and COPR

for a POC what's a good way to hack it? just have a cheat and use the date as a counter temporarily maybe

yeah

3 mins until some of us need to run to our next meeting

the goal is to have a way to tell at a glance how many times something was built of a single source, but for PoC stuff seeing the no-source rebuilds work it's easy to do with a sufficiently constrained timestamp

I'm going to try to have a script that shows how the counter mechanism should work

It'll be in a repo on codeberg: https://codeberg.org/Conan_Kudo/mock-nochange-rebuild

anyway that's it for me

is there anything else from anyone?

if not... Davide Cavalca?

nothing else from me, we can close