centos-hyperscale-sig
LOGS
<@davide:cavalca.name>
16:01:14
!startmeeting CentOS Hyperscale SIG
<@meetbot:fedora.im>
16:01:16
Meeting started at 2025-11-19 16:01:14 UTC
<@meetbot:fedora.im>
16:01:17
The Meeting name is 'CentOS Hyperscale SIG'
<@davide:cavalca.name>
16:01:22
morning folks
<@davide:cavalca.name>
16:01:25
!topic Roll call
<@conan_kudo:matrix.org>
16:02:32
!hi
<@zodbot:fedora.im>
16:02:35
Neal Gompa (ngompa) - he / him / his
<@rcolebaugh:matrix.org>
16:02:51
!hi
<@zodbot:fedora.im>
16:02:54
Raymond Colebaugh (rcolebaugh) - he / him / his
<@salimma:fedora.im>
16:03:13
!hi
<@zodbot:fedora.im>
16:03:14
Michel Lind (salimma) - he / him / his
<@james:reilly.asia>
16:04:05
!hi
<@zodbot:fedora.im>
16:04:13
James Reilly (jreilly1821) - he / him / his
<@davide:cavalca.name>
16:05:30
let's get started
<@davide:cavalca.name>
16:05:32
!topic Followups
<@davide:cavalca.name>
16:05:39
any follwups from the last meeting?
<@salimma:fedora.im>
16:08:51
been busy with internal stuff recently and have not had the time to do much
<@davide:cavalca.name>
16:09:11
yeah for context on the fb side we've been in roadmapping hell for the last week or so
<@davide:cavalca.name>
16:09:18
so not much time for actual work unfortunately
<@salimma:fedora.im>
16:09:41
no issue with the wprof we built in hsx at least. I'm trying to find time to unblock its dependencies in Fedora so we can move it there
<@salimma:fedora.im>
16:10:21
I have something else for open floor later that's related to some roadmap discussions
<@conan_kudo:matrix.org>
16:11:24
things kept coming up that kept me from finishing the kernel rebase
<@conan_kudo:matrix.org>
16:11:30
hopefully will get it done this week
<@salimma:fedora.im>
16:12:09
we need to figure out a game plan for the kernel. not just a Hyperscale issue, but centos and fedora too -- it's really really hard to contribute to that package
<@salimma:fedora.im>
16:12:41
or even to maintain some of the parts separately (e.g. the perf standalone tarball has been broken for at least half a year)
<@salimma:fedora.im>
16:13:27
but without some examples of how to do things differently we can't really get package maintainers to budge as they understandably tend to be risk averse
<@salimma:fedora.im>
16:13:35
hmm, worth a centos connect / fosdem talk proposal?
<@conan_kudo:matrix.org>
16:13:44
I think it's worth a connect talk
<@conan_kudo:matrix.org>
16:13:55
you want to propose something we can do together?
<@salimma:fedora.im>
16:13:57
so I don't keep talking about my tools, which I have not had time to work on this year :P
<@salimma:fedora.im>
16:14:10
yeah, I'll propose something
<@conan_kudo:matrix.org>
16:14:29
I'm kind of frustrated too, because the ARK is simultaneously easy and painful
<@conan_kudo:matrix.org>
16:14:35
I'm kind of frustrated too, because the ARK workflow is simultaneously easy and painful
<@conan_kudo:matrix.org>
16:14:53
I don't love how slow git is because the kernel repo is enormous
<@conan_kudo:matrix.org>
16:15:14
it reminds me how much I don't like merged-source / source-git stuff
<@salimma:fedora.im>
16:15:17
nuclear option - do we have to use ARK?
<@salimma:fedora.im>
16:15:34
are there other RPM based distros that maintain their spec in a nicer way
<@conan_kudo:matrix.org>
16:15:35
if I had dedicated time to pull all the packaging out of ARK, I would
<@conan_kudo:matrix.org>
16:15:59
most RPM distros don't do it like this
<@salimma:fedora.im>
16:16:09
the spec is a mess, I wish there's a way to toggle "build only the kernel" or "build everything but the kernel" e.g. the old kernel-tools
<@salimma:fedora.im>
16:16:35
that will make it much easier to keep rebasing on older EL releases, where normally it's the userspace tools that no longer compile
<@conan_kudo:matrix.org>
16:16:40
if RPM wasn't so old in EL, I'd suggest we actually look at rewriting the whole spec to use newer features
<@conan_kudo:matrix.org>
16:16:48
it would make the spec so much nicer if we could
<@salimma:fedora.im>
16:17:12
if we can get a proof of concept working, I think it's worth doing for Fedora/ELN?
<@salimma:fedora.im>
16:17:22
it will just take longer to benefit from it in Hyperscale
<@conan_kudo:matrix.org>
16:17:26
yeah
<@salimma:fedora.im>
16:17:29
because no we're not rebasing RPM\
<@salimma:fedora.im>
16:17:43
it will also be hell for Meta because that means we have to port RPM CoW urgently
<@salimma:fedora.im>
16:17:52
because no we're not rebasing RPM
<@conan_kudo:matrix.org>
16:17:55
well that needs to be done in 2026 anyway
<@salimma:fedora.im>
16:18:02
anyway, we shouldn't hog this for the kernel talk
<@conan_kudo:matrix.org>
16:18:24
since CentOS Stream 11 will arrive with Fedora 46
<@conan_kudo:matrix.org>
16:18:26
in 2027
<@salimma:fedora.im>
16:19:04
yeah. still trying to find someone to sacrifice to do that work :/
<@salimma:fedora.im>
16:19:47
it's thankless since there's no clear path to upstream. like the openssh problem but bigger
<@conan_kudo:matrix.org>
16:20:37
we need to organize some kind of in-person meeting with the rpm folks
<@conan_kudo:matrix.org>
16:20:47
the async back and forth isn't working
<@salimma:fedora.im>
16:21:35
well that's the "no viable path forward" too
<@salimma:fedora.im>
16:21:43
unless we organize the in person meetup in Finland
<@salimma:fedora.im>
16:21:53
we *have* had in person meetups... twice. in Brno and Boston
<@salimma:fedora.im>
16:22:02
but the main blocker is not in the room
<@conan_kudo:matrix.org>
16:22:45
mmm
<@conan_kudo:matrix.org>
16:22:51
I have ideas but let's take that offline
<@davide:cavalca.name>
16:24:18
good segway to
<@davide:cavalca.name>
16:24:19
!topic Announcements
<@davide:cavalca.name>
16:24:49
a bunch of us will be at centos connect and fosdem next year
<@davide:cavalca.name>
16:27:51
anything else for announcements?
<@salimma:fedora.im>
16:31:53
I take it no
<@conan_kudo:matrix.org>
16:33:04
nope
<@davide:cavalca.name>
16:35:38
!topic Tickets
<@davide:cavalca.name>
16:37:04
I don't think we have anything here
<@davide:cavalca.name>
16:37:11
!topic Membership
<@conan_kudo:matrix.org>
16:38:34
nothing here either, I think?
<@davide:cavalca.name>
16:38:59
yep
<@davide:cavalca.name>
16:39:05
!topic Miscellaneous
<@davide:cavalca.name>
16:39:13
I do have one thing here
<@davide:cavalca.name>
16:39:35
we have the hangout today, but it's 2pm PT, which is too late for Michel to attend
<@davide:cavalca.name>
16:39:48
we should see if we can reschedule to a better time
<@davide:cavalca.name>
16:40:02
(not this specific instance necessarily, I mean in general)
<@salimma:fedora.im>
16:40:43
yeah, too late for today in any case I guess
<@salimma:fedora.im>
16:41:03
is there a good alternative to doodle for finding a time? iirc framasoft used to have one
<@conan_kudo:matrix.org>
16:42:05
I use whenisgood because it's the only one that lets you set your timezone when you're filling it out
<@salimma:fedora.im>
16:42:56
yeah whenisgood is the other one I was thinking of, but ugh we're in 2025 and there's no good open source solution for this huh
<@salimma:fedora.im>
16:43:25
this is framapiaf's solution - https://framadate.org/abc/en/
<@salimma:fedora.im>
16:43:41
i will check the guided tour after this and see if it let you see times in local tz
<@conan_kudo:matrix.org>
16:43:44
yeah but you cannot have it show it in your timezone when selecting
<@conan_kudo:matrix.org>
16:43:51
discovered that with a kde meeting last month
<@salimma:fedora.im>
16:44:26
oh, KDE tried framadate already?
<@conan_kudo:matrix.org>
16:44:45
yeah
<@salimma:fedora.im>
16:44:45
(I guess shoehorning a time for jameswill be super hard, sorry James)
<@conan_kudo:matrix.org>
16:44:55
we could actually start doing two
<@conan_kudo:matrix.org>
16:45:13
the spread of timezones is a bit much to fit into one, but we could create two at different times
<@conan_kudo:matrix.org>
16:45:34
the constraint is really Davide and I figuring out good second times
<@salimma:fedora.im>
16:46:45
oh amazing the time is wrong on calendar.fedoraproject.org again of course
<@davide:cavalca.name>
16:46:56
I also don't mind making this floating if we want to try that
<@salimma:fedora.im>
16:47:01
it had Nov 5 since fedocal can't do "nth week of the month"
<@salimma:fedora.im>
16:47:05
all our tooling sucks
<@davide:cavalca.name>
16:47:07
I do like the chance to chitchat so it'd be nice to keep this going if we can
<@salimma:fedora.im>
16:47:17
floating how?
<@davide:cavalca.name>
16:47:32
different times on different months to better cater to different tzs
<@salimma:fedora.im>
16:48:26
oh right, yes
<@conan_kudo:matrix.org>
16:48:38
well folks, I gotta go, have to go to the dentist
<@salimma:fedora.im>
16:48:43
good luck
<@salimma:fedora.im>
16:48:56
!action Michel will set up a poll to find meeting times
<@salimma:fedora.im>
16:49:25
are we sticking to "nth week of the month" (with the caveat it will make it require manual interventions on fedocal) or are we going to try to make it "every 4 weeks" ?
<@salimma:fedora.im>
16:49:42
I think "nth week of the month" is fine, I just need a todo for myself to keep checking fedocal for when they diverge
<@davide:cavalca.name>
16:51:26
I don't care either way as long as we're consistent
<@salimma:fedora.im>
16:53:16
moving on..
<@salimma:fedora.im>
16:53:54
oh but I will edit fedocal after this to make sure people who subscribed to the hyperscale calendar can find the meegin
<@davide:cavalca.name>
16:56:54
I think that's all for this week
<@davide:cavalca.name>
16:57:07
thanks everyone, and see you soon
<@davide:cavalca.name>
16:57:13
!endmeeting