16:01:00 <dcavalca> #startmeeting Hyperscale SIG
16:01:00 <centbot> Meeting started Wed Oct 13 16:01:00 2021 UTC.  The chair is dcavalca. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:00 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:01:08 <dcavalca> #chair dcavalca jvreeland
16:01:08 <centbot> Current chairs: dcavalca jvreeland
16:01:13 <dcavalca> morning folks
16:01:20 <jvreeland> morning
16:02:13 <dcavalca> let's get started
16:02:15 <dcavalca> #topic Followups
16:02:41 <dcavalca> I spent some time tidying up the user docs
16:02:52 <dcavalca> https://sigs.centos.org/hyperscale/ should be mostly usable now
16:03:44 <dcavalca> the only other thing on my end is figuring out what to do for openssl 3.0, which I've made zero progress on due to lack of time so far
16:04:08 <dcavalca> anything else for followups?
16:04:28 <jvreeland> I've been rebuilding packages for aarch64, into the ls now hoping to finish up this week or next
16:04:50 <bkircher> Hey there
16:05:01 <dcavalca> thanks jvreeland
16:05:05 <dcavalca> hey bkircher
16:05:09 <jvreeland> I've also been working on the kernel rebase which will hopefull be done soon just issues with tools
16:05:22 <Eighth_Doctor> we should have libtraceevent soon in epel8
16:05:36 <dcavalca> yeah, I saw the update for that went through
16:05:46 <dcavalca> so we should be unblocked there
16:05:47 <Eighth_Doctor> https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2021-83b30f74f4
16:06:05 <Eighth_Doctor> it should push out in tonight's compose
16:06:40 <Eighth_Doctor> that should let us rebase to the RHEL9 kernel tree for our kernel
16:07:34 <dcavalca> perfect
16:07:58 <jvreeland> sounds good
16:08:01 <Eighth_Doctor> we should really start identifying stuff to pull into their kernel
16:08:20 <Eighth_Doctor> https://pagure.io/centos-sig-hyperscale/sig/issue/72
16:10:07 <dcavalca> yep
16:10:21 <dcavalca> moving on to
16:10:22 <dcavalca> #topic Announcements
16:10:42 <dcavalca> #info Hyperscale Dojo talk is out: https://www.youtube.com/watch?v=s6unjYyjbXk
16:10:46 <dcavalca> this went very well
16:11:06 <dcavalca> #info 2021Q3 Hyperscale SIG report is out: https://blog.centos.org/2021/10/centos-hyperscale-sig-quarterly-report-for-2021q3/
16:11:13 <dcavalca> thanks everyone for the help there
16:11:36 <osandov> hooray
16:11:52 <dcavalca> the only other item I wanted to discuss here is iptables in c9s
16:12:13 <dcavalca> RH isn't interested in shipping or supporting iptables-legacy: https://bugzilla.redhat.com/show_bug.cgi?id=2011531
16:12:25 <Eighth_Doctor> oh?
16:12:27 <dcavalca> so for the time being, we only have the nft backend and we don't have libiptc, which breaks the systemd build
16:12:27 <Eighth_Doctor> what about it?
16:12:47 <dcavalca> (our systemd build, in the official one they've just ripped it out)
16:13:10 <Eighth_Doctor> I thought systemd was moving to eBPF for this stuff?
16:13:14 <Eighth_Doctor> I know libebpf is in RHEL 9
16:13:27 <dcavalca> Eighth_Doctor: I'm trying to figure this out as well and am asking around
16:13:47 <dcavalca> but if we do end up having to keep supporting -legacy, I suspect the easiest will be to get it packaged in EPEL
16:13:57 <dcavalca> as it's nearly selfcontained in a bunch of packages
16:14:51 <Eighth_Doctor> yeah, I think this won't be that bad
16:14:59 <Eighth_Doctor> doing a build in EPEL will be kind of a blessing, actually
16:15:19 <Eighth_Doctor> because it'll allow third-party firewalls to work on RHEL 9 (like shorewall)
16:15:34 <dcavalca> jvreeland: out of curiosity, what do y'all do for iptables there? are you using nft?
16:16:27 <jvreeland> We use iptables but before the sig I was already telling people we'd switch off when 8 came out. I imagine we're less tied to it than you are.
16:17:01 <dcavalca> yeah well, right now for me the issue is that we don't even have nft enabled in our kernels
16:17:28 <dcavalca> and the last time  I had that conversation internally (~ 2 years ago) folks didn't seem terribly interested in supporting it
16:17:38 <dcavalca> so, we'll see how this goes
16:17:53 <Eighth_Doctor> this reminded me to start that conversation internally too
16:18:00 <dcavalca> I would personally love to see everything move to a shiny new BPF thing, but I haven't seen anything remotely close to production use for that yet
16:18:27 <dcavalca> anybody else has any announcementsd?
16:18:28 <Eighth_Doctor> I have a nasty feeling that we're relying on it somewhere, but I think our reliance is a bit weak
16:19:30 <Eighth_Doctor> nothing right now
16:19:46 <Eighth_Doctor> I'm waiting for a new c9s based hyperscale kernel for CBS
16:19:59 <Eighth_Doctor> so that we can start getting things going with arrfab
16:20:28 <dcavalca> alright, time for
16:20:31 <dcavalca> #topic Tickets
16:20:44 <dcavalca> bkircher: any update on qemu for #67?
16:20:56 <bkircher> hate to say it but no
16:21:10 <bkircher> but seems I got a bit of time for this next week
16:21:33 <Eighth_Doctor> I did discover we've got a problem for c9s qemu in EPEL
16:21:43 <Eighth_Doctor> libiscsi-devel is missing in CRB for c9s
16:21:58 <bkircher> yep, seen this
16:22:05 <dcavalca> didn't we have a ticket for that already?
16:22:26 <dcavalca> https://bugzilla.redhat.com/show_bug.cgi?id=2004762 yeah
16:23:11 <Eighth_Doctor> yep, I filed it for an unrelated project, and then I found it out it affects us too
16:24:03 <dcavalca> given that this is impacting EPEL now, we should escalate it in the WG meeting this afternoon
16:24:17 <dcavalca> which I'll miss as it conflicts with the board meeting, so if someone can do that it'd be great
16:25:34 <dcavalca> for #58, it looks like we'll be getting source repos for c9s, so it should sort itself out there
16:25:47 <dcavalca> we can probably leave it open for tracking, but I don't think there's anything actionable on our end for the time being
16:26:04 <dcavalca> we already talked about the kernel
16:26:10 <dcavalca> did we have any other tickets to discuss here?
16:26:48 <Eighth_Doctor> hmm
16:27:14 <Eighth_Doctor> what's with the branch build tickets still open?
16:27:38 <dcavalca> which one's that?
16:28:21 <jvreeland> #65 and #12?
16:28:33 <jvreeland> speaking of iptables
16:28:42 <dcavalca> #65 I haven't gotten around yet
16:28:58 <dcavalca> it's something we've forked internally that I need to clean up to get upstreamed
16:29:25 <dcavalca> #12 I'll check, I think most of it was done but I don't remember if we actually pushed the iptables package or not
16:29:38 <dcavalca> but good call, I'll take an action on this
16:29:54 <dcavalca> #action dcavalca check on stale tickets for packages to branch
16:30:04 <Eighth_Doctor> there's also #64
16:30:16 <Eighth_Doctor> but yeah, there's a bunch of this
16:30:31 <Eighth_Doctor> I wonder if we should have a tag for those to easily track them
16:31:13 <dcavalca> yeah that's a good idea
16:33:04 <dcavalca> moving on
16:33:08 <dcavalca> #topic Membership
16:33:18 <dcavalca> I don't think we have anything here this week
16:33:47 <dcavalca> #topic Misc
16:33:53 <dcavalca> anything else folks wanna talk about?
16:34:50 <Eighth_Doctor> I'd like for us to ramp up and have the ability to produce the c9s based spins sooner rather than later :)
16:35:36 <Eighth_Doctor> at this point, I think the main blocker is the kernel
16:35:37 <jvreeland> Building the kernels in mock as we speak
16:35:46 <Eighth_Doctor> but I've started looking at the userspace stuff
16:36:05 <Eighth_Doctor> we probably can start with just building the latest c9s kernel with hyperscale modifications to the config
16:36:14 <Eighth_Doctor> and then start stacking on top
16:38:44 <Eighth_Doctor> in theory, epel9-next should be open now? carlwgeorge or tdawson may know
16:38:53 <Eighth_Doctor> if it is, we should start doing that stuff
16:38:57 <dcavalca> Eighth_Doctor: once we have the kernel, I think that's the last stopper to testing out a dedicated builder in CBS too?
16:39:04 <Eighth_Doctor> yes
16:39:15 <dcavalca> oh I didn't realize epel9-next was open already
16:39:22 <dcavalca> I saw the repos but I thought it was still being worked on
16:39:40 <Eighth_Doctor> I don't know for certain, but I think it is available now
16:40:39 <dcavalca> alright, something else to followup for this afternoon if someone can make the WG meeting
16:41:07 <carlwgeorge> it's not available yet
16:41:39 <carlwgeorge> epel-release failed to build, need to rework some of the koji tag inheritance
16:43:58 <dcavalca> thanks for the update carlwgeorge
16:46:46 <dcavalca> looks like we'll end early today
16:46:55 <dcavalca> thanks everyone!
16:47:07 <dcavalca> #endmeeting