Meeting started at 2026-02-25 16:02:01 UTC

The Meeting name is 'CentOS Hyperscale SIG'

a bit surprised there was nobody using this since fesco yesterday, I guess council was not meeting today?

!hi

Michel Lind (salimma) - he / him / his

!hi

Neal Gompa (ngompa) - he / him / his

Council is doing a video meeting right now

it's just finishing up now

Jef Spaleta has decided to restart the periodic video meetings that mattdm stopped doing a while back

!hi

Pawel Zmarzly (pawo2500) - he / him / his

Davide asked me to start the meeting, not sure if he's joining late or not joining

but we can wait until ... let's say 3 more minutes for roll call

sounds good

alright, let's start then

just in case this is the last meeting summary https://meetbot.fedoraproject.org/meeting-2_matrix_fedoraproject-org/2026-02-11/centos-hyperscale-sig.2026-02-11-16.08.html

and I think Conan Kudohas some updates re: talking to different stakeholders about enabling btrfs in ARK?

we did a bad job with the bang commands for that

yeah, no info or anything else :P

yeah, a bit

So the last Fedora ELN SIG meeting resulted in yselkowitz and I conversing on a path to upstreaming Hyperscale work into Fedora to enable Fedora ELN Hyperscale

so the action item for Davide Cavalca to file a Jira to talk to the RHEL kernel folks and ARK is now slightly more important 😅

There's now a parent ticket about Fedora ELN Hyperscale to track all work on that front

so over the coming weeks, I'm going to start upstreaming our conditionals for CentOS Hyperscale spin components into Fedora

and also create an eln-hyperscale release package in Fedora too

this will be required for Anaconda stuff to work

oh... interesting.

this will be built only for eln, not for rawhide, right?

yeah

cc pzmarzlyin case you end up doing some ELN integration work

normal rawhide would already have most of what we do

the one discussion I noticed in that ELN meeting that's relevant to us is Yaakov asking what to do if different SIGs end up needing the same package not in core ELN

I'm not sure exactly how to deduplicate this safely but yet ensuring that, say, that eln-hyperscale release package you are working on will ensure we still have the common packages but not too much extra :)

!hi

Davide Cavalca (dcavalca) - he / him / his

oh we summoned Davide :)

let's give this topic a few mins extra in case people have follow-ups, then I guess we can discuss the membership cleanup, the other perennial follow-up :)

Sorry running late today

I'll get the ticket sorted out today

ok, that's it for kernel I guess. pzmarzlyany news on membership cleanup, iirc last time the script was not taking into account MR reviews?

correct. i changed the script a bit, now it scrapes much more data from gitlab

(https://inactivity-monitoring-8e4c02.gitlab.io/)

huh we are allowed to have that subdomain?

hmm gitlab pages are enabled on gitlab.com/CentOS org

and i got this domain automatically for that repo

so i guess yes?

ah interesting, I wonder if we can have 'hyperscale' show up as part of the domain

a bit surprising it ignores namespacing completely but maybe we can customize it

hmm i think maybe centos org admins can, but I don't see an option

we can always use CNAME if we wanted

oh, last time we had issues with people being added manually too, right? so no corresponding FAS

but i don't think we need a fancy domain for that?

its mostly for us

I think we can make a decision today about what to do in that situation.

Proposal: if someone has no FAS linked we remove them first, they can get re-added later

yes, but that turned out not to be an issue in the end

my script was just pulling people through inheritance

yeah, I'm just worried people don't know what sort of inactivity monitor it is, the curret subdomain is a bit generic

so i changed it to first pull FAS, then try finding GitLab

so i changed it to first pull list from FAS, then try finding GitLab

whereas before I was using GitLab list and trying to get FAS users out of that

i can add a link to the repo? at the top of the page?

I think once you click through it's obvious though, the page title says hyperscale

but yeah linking to the source is always good

we can consider adding this to the SIG docs too

I like how this also flags the repos that has had no activity, this will help cleaning up things we don't need anymore

now, we still have a problem if we have someone in FAS group but they have no GitLab account linked

but that's a good signal of inactivity tbh

I wonder if there's a way to identify archived repos, now that we're on gitlab, so once we archive a repo it does not show up again in the report

agreed

if you have context on these repos, wdyt about meeting in person and going through the list?

good idea

how do i use bot to add action items?

(go back up to when the meeting starts, or you can ask the bot for help)

!help

hmm the help command is not discoverable it seems :P

looking great, thanks

salimma gave a cookie to pawo2500. They now have 1 cookie, 1 of which was obtained in the Fedora 43 release cycle

first cookie!

(this is like our internal tool but you're rate limited to one per person per Fedora release cycle)

oh, that's quite slow grind

oh one more request - can we identify when someone joins the SIG?

if you just joined and have no activity yet and get culled that's a bit rough :P

the script pulls the list from FAS group membership

ah, but yes, you will be red

hmm i will check if FAS exposes timestamps

yeah, I can see at least one person I know should be excluded for that reason

but most of the ones in red are safe to remove

ok... let me look at my cheatsheet to see what we need to discuss next

anyone has something to announce here?

going once..

going twice..

I guess nothing much has happened, I know I'm mostly busy with the Fedora release cycle right now :P

oh

how is your AI-powered linter going?

how is your AI-powered linter for specfiles going?

eh is that something I want to announce here :P

well, maybe lets first see if it can work :D

well, maybe lets first see if it can even work :D

and yeah, if it does then the audience is bigger than hyperscale anyways

something like Debian Janitor would be super awesome but that's probably not Hyperscale specific

right

so maybe i shouldn't have brought it here

eh, it's tangentially relevant and we don't have any real announcement :)

remind me - did we move things to GitLab already?

Yep

should we delete pagure repos and see if anything breaks?

Though there's no way to mark the pagure stuff as archived

yeah, pagure not having archive status is a bit unfortunate

I guess we can update the README

I see some other pagure issue trackers that got around it by making sure the latest filed issue is the migration one

so we can do that, it's easier than ... oh yeah we don't have a README for the sig racker since that has no repo.

iS SN# the old number from Pagure? what does SN stand for

i think it is ID from pagure? if not then its pagure-importer's internal ID, but that would be a bit useless

that does look like the issue ID

yeah looks correct to me

oh yeah spot checkign one issue, that's th pagure id

picking a random issue -- let's see : https://gitlab.com/CentOS/Hyperscale/tracker/-/issues/8

i just noticed that cross-issue references are broken

Conan Kudodo we even want to fix this? (erofs in EL missing zstd support)

if so can't we just have our own erofs build

in our own kernel I mean

for example, https://gitlab.com/CentOS/Hyperscale/tracker/-/issues/115 says `#11` in it

and it should point to SN#11, but it points to GL#11. Ehhhh

yeah

that's to be expected I guess :(

oh well, one off pain

alright, if nobody has an actionable ticket let's move on to

nobody is in the queue I guess, and we already noted action items for doing the cleanup

'airing of grievances' :P

lol

https://www.youtube.com/watch?v=ATXV3DzKv68

oh wow there are updated versions

if there's nothing else, I guess those of us who want to / need to attend the FRCL meeting gets to have a few minutes' break

thanks all for coming and for the discussion and contributions!