15:04:09 <davdunc> #startmeeting fedora_cloud_meeting
15:04:09 <zodbot> Meeting started Thu Aug 31 15:04:09 2023 UTC.
15:04:09 <zodbot> This meeting is logged and archived in a public location.
15:04:09 <zodbot> The chair is davdunc. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
15:04:09 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:04:09 <zodbot> The meeting name has been set to 'fedora_cloud_meeting'
15:04:18 <davdunc> #topic roll call
15:05:22 <mhayden> .hello2
15:05:23 <zodbot> mhayden: mhayden 'Major Hayden' <mhayden@redhat.com>
15:05:29 <davdunc> #chair mhayden
15:05:29 <zodbot> Current chairs: davdunc mhayden
15:05:34 <nzwulfin> .hello
15:05:34 <zodbot> nzwulfin: (hello <an alias, 1 argument>) -- Alias for "hellomynameis $1".
15:05:42 <VSokolenko> guys, how may I join the call here?
15:05:45 <davdunc> #chair nzwulfin
15:05:45 <zodbot> Current chairs: davdunc mhayden nzwulfin
15:05:47 <nzwulfin> .hello nzwulfin
15:05:47 <VSokolenko> .hello 4sokol
15:05:47 <zodbot> nzwulfin: Sorry, but user 'nzwulfin' does not exist
15:05:50 <zodbot> VSokolenko: 4sokol 'Vladimir Sokolenko' <vladimir@sokolenko.org>
15:06:04 <davdunc> #chair 4sokol
15:06:04 <zodbot> Current chairs: 4sokol davdunc mhayden nzwulfin
15:06:07 <nzwulfin> one of these days i'll remember my FAS the first try
15:06:26 <davdunc> VSokolenko you are in!
15:06:32 <nzwulfin> .hello mmicene
15:06:33 <zodbot> nzwulfin: mmicene 'Matt Micene' <nzwulfin@gmail.com>
15:07:18 <davdunc> excellent.
15:08:06 <davdunc> #topic actions from last meeting
15:08:34 <davdunc> #action davdunc +friends hackfest on Azure Image support
15:09:03 <davdunc> mhayden: it seems like this is something we should be able to do with os-build now, yes?
15:09:14 <davdunc> or are we still awaiting the btrfs support.
15:09:39 <mhayden> we are still waiting on btrfs support
15:09:51 <mhayden> however, there is a lot of interim work that can be done in the meantime to get things wired up
15:10:12 <davdunc> yep. we can do it with kiwi now and we need that for the WSL as well.
15:11:00 <mhayden> will kiwi handle the upload/import at providers?
15:11:02 <davdunc> let's put some time on the schedule in the next week for the config.
15:11:41 <davdunc> no. We still need work there.
15:12:16 <mhayden> it sounds like pbrobinson has some knowledge here and has given me some tips on things to try
15:12:27 <mhayden> we might be able to send him some cookies in exchange for help 😉
15:12:35 <davdunc> :) sounds good.
15:13:10 <davdunc> I spoke to him the other day about adding composable builds and the btrfs thing.
15:14:17 <davdunc> if that's what gets us a repeatable upload to Azure before F39 release, cookies are just a start!
15:15:47 <davdunc> it sounded like you had some knowledge here that we want to capture as well.
15:15:55 <davdunc> you == mhayden
15:15:58 <davdunc> :)
15:16:14 <mhayden> hah yes, i've done it with image builder myself in Fedora several times, and we have osbuild in koji already
15:16:21 <mhayden> Fedora Azure images, i mean
15:16:39 <davdunc> perfect. Can we just turn it on for the release?
15:17:04 <davdunc> or are you building a different image?
15:17:10 <mhayden> we _could_ generate an Azure-compatible image with osbuild today (ondrej put one in last week) but we're missing btrfs
15:17:14 <mhayden> everything else is fine
15:17:46 <mhayden> pbrobinson said that pungi would poke koji to make the build, but i doubt we can get that in before F39 goes out
15:17:51 <mhayden> unless you know something differently
15:17:58 * mhayden is in two meetings simultaneously
15:18:53 <davdunc> well. . . I think that's a way to do it, but it's also a way for the btrfs support to be de-prioritized.
15:19:55 <mhayden> true. perhaps this is a good thread for the ML and bring in some osbuild folks on that
15:20:06 <davdunc> It seems like we can handle this with the Azure collection in ansible with the btrfs in-tact.
15:21:05 <davdunc> agreed.
15:21:59 <davdunc> I think the ML is a great place to determine what our timeline is on the support.
15:22:41 <mhayden> okay, i can start a thread
15:22:52 <davdunc> super.
15:24:37 <davdunc> Alright.
15:25:25 <davdunc> * upload and test images for Azure Marketplace
15:25:52 <davdunc> i have images for testing with btrfs in my account, but haven't moved them to the Fedora account.
15:26:05 <davdunc> with btrfs, we are good
15:26:57 <davdunc> I am not testing nightlies at the moment though.
15:27:35 <pbrobinson> I believe also that osbuild has some form of cloud push functionality but I'm far from expert on those bits
15:27:37 <mhayden> 👍
15:27:52 <davdunc> definitely does pbrobinson
15:28:13 <davdunc> it's pretty sophisticated in the way that it handles the azure uploads from what I understand
15:28:21 <pbrobinson> I am actually today looking at some partition related things in osbuild so I'm practically the expert there now (LOL!!!)
15:28:30 <davdunc> haha.
15:28:34 <pbrobinson> yes, and it's what RHEL uses
15:28:40 <mhayden> yes, i use osbuild-composer to push directly to cloud and it's kinda nifty
15:29:23 <davdunc> well, even if we don't build the image in the os-build, can't we just push it to Azure using the composer?
15:29:37 * davdunc thinks this is the most important topic of the week.
15:30:00 <davdunc> I want to get this resolved and on the happy path for F39.
15:30:13 <dustymabe> heyo
15:30:20 <davdunc> there he is!
15:30:28 <davdunc> #chair dustymabe
15:30:28 <zodbot> Current chairs: 4sokol davdunc dustymabe mhayden nzwulfin
15:30:36 * dustymabe also in two meetings - grab my attention if there's something that needs it!
15:31:10 <davdunc> dustymabe: all good. Understand that we can
15:31:38 <davdunc> can't all be where we want to be all the time. :) namely, "in one meeting at a time"
15:32:07 <davdunc> okay, so . . .
15:32:50 <davdunc> #action davdunc to work with mhayden to determine if we can use the composer separately from the image build for the Azure uploads
15:33:53 <davdunc> same thing is applicable to OCI
15:34:25 <davdunc> #action davdunc to work with mhayden to determine if we can use the composer separately from the image build for the OCI uploads.
15:34:31 <mhayden> i believe osbuild/osbuild-composer handles builds/uploads there, but i'd have to double check
15:34:42 <mhayden> our QCOW2 that we publish in the compose works well on OCI
15:34:42 <davdunc> it doesn't handle btrfs.
15:34:46 <mhayden> (without modifications)
15:34:59 <mhayden> right, no btrfs
15:34:59 <davdunc> let's keep our build system if we can until we have btrfs support.
15:35:11 <mhayden> but the qcows we already make are just fine for OCI
15:35:31 <davdunc> yea. they use cloud-config so we are set there.
15:35:51 <dustymabe> osbuild/IB doesn't support btrfs because?
15:35:51 <mhayden> OCI is pretty forgiving on the formats they'll take last time i looked at them
15:36:10 <mhayden> dustymabe: just hasn't been prioritized yet since RHEL doesn't have btrfs
15:36:28 <mhayden> but i think someone on the osbuild team has enumerated what needs to be completed
15:36:30 <davdunc> dustymabe: yes.
15:36:38 <dustymabe> https://github.com/osbuild/osbuild/blob/main/stages/org.osbuild.mkfs.btrfs exists ?
15:36:43 <davdunc> yes to both statements.
15:36:55 <mhayden> i was told there's a bug in there to fix
15:36:57 <davdunc> it does, but it's not fully functional.
15:37:17 <dustymabe> ahh - open issue somewhere maybe?
15:37:30 <dustymabe> "fix a bug" is a much lower bar than "implement it from scratch"
15:37:37 <davdunc> #link https://github.com/osbuild/osbuild/blob/main/stages/org.osbuild.mkfs.btrfs
15:37:39 <mhayden> might be this: https://github.com/osbuild/osbuild/issues/517
15:37:53 <dustymabe> isn't there an open issue to have workstation use IB for image building? doesn't workstation use btrfs?
15:38:08 <mhayden> let's get an email thread going with a few folks from the osbuild folks and get their input
15:38:12 <davdunc> #link https://github.com/osbuild/osbuild/issues/517
15:38:15 <mhayden> might be able to influence their timelines 😜
15:39:42 <davdunc> yea I think that's the best idea there. I am hoping that someone from Meta will be able to jump on anything related to btrfs directly.
15:40:39 <davdunc> it would be great to have parity. pbrobinson any roadmap on WSL in the tool?
15:41:44 <davdunc> alright. that's probably enough to keep us rolling.
15:43:03 <davdunc> That takes care of two issues in our meetings issues.
15:43:18 <davdunc> #link https://pagure.io/cloud-sig/issue/400
15:43:26 <davdunc> #link https://pagure.io/cloud-sig/issue/399
15:44:19 <davdunc> #topic Provide unversioned cloud images
15:44:32 <davdunc> #link https://pagure.io/cloud-sig/issue/396
15:44:57 <davdunc> What doe you think of a "latest" version?
15:45:17 <davdunc> gonna give this topic 4 minutes at most .
15:45:47 <mhayden> hmm, that's an interesting request, but i don't know how that would work at a cloud provider
15:46:23 <davdunc> I think that having a "latest" is something that just happens on GCP and then Azure Marketplace is the same.
15:46:26 <mhayden> i mean, i like the way it works for containers where i can get quay.io/fedora/fedora:latest
15:46:45 <dustymabe> kubevirt works like that too :)
15:46:58 <dustymabe> quay.io/fedora/fedora-coreos-kubevirt:stable
15:47:03 <davdunc> the name changes though and I think that the hope was that we could make a single search work.
15:47:08 <mhayden> this one gets tricky since each cloud provider has support for certain things
15:47:19 <davdunc> exactly.
15:47:32 <pbrobinson> davdunc: what's WSL?
15:47:35 <davdunc> on Amazon EC2, there are AMI public parameters.
15:47:48 <davdunc> WSL = Windows Subsystem for Linux
15:48:36 <davdunc> re: Latest, it could also be a qcow we have in the downloads that always provides the latest release.
15:48:58 <davdunc> as right now, we don't have a naming convention that is just the most recent.
15:49:16 <davdunc> in that case, it could just be a symlink to the file.
15:49:29 <pbrobinson> looks like someone has already done WSL support, there's at least a WSL image type
15:49:59 <davdunc> we want an official one. It's not one we build
15:50:08 <pbrobinson> https://github.com/osbuild/images/commit/d7bcb358b0cd33a7b9957ed35fbe93ede676daf5
15:50:14 <mhayden> got bumped by the web chat 😭
15:50:43 <davdunc> LOL! Awesome!
15:51:33 <davdunc> I think it's a big help to have that consistency. I know that you don't have to use it pbrobinson, but I would like to use it and it's not always functional today.
15:52:11 <davdunc> sometimes i have to windoze.
15:52:14 <pbrobinson> davdunc: I've used it once on arm, but I wanted to check the acronym before I commented
15:52:22 <davdunc> alright.
15:52:30 <dustymabe> shouldn't the WSL image be more akin to the Fedora Container image versus the Fedora Cloud image ?
15:52:34 <pbrobinson> I think the more ways we can allow Fedora to be used, what ever the OS, the better
15:52:56 <davdunc> pbrobinson: +1.
15:53:20 <pbrobinson> dustymabe: probably, I didn't look at how it was crafted in that commit, just it was there (I saw it earlier digging through other stuff)
15:53:39 <davdunc> dustymabe: the container build should be consistent with the cloud image too.
15:53:55 <dustymabe> davdunc: hmm
15:54:10 <dustymabe> not in my opinion - for example a container doesn't need cloud-init
15:54:43 <davdunc> right or other tools.
15:54:53 <davdunc> I see what you are driving at.
15:55:03 <dustymabe> that's why I was asking :) - just trying to understand what the goal was
15:55:59 <davdunc> yea. WSL doesn't need cloud-init either, but as a team effort, it's a consistent fit with the Cloud Sig goals.
15:56:39 <davdunc> more discussion needed on the latest, but it's something we should look at writing a change proposal to take a look at it in a realistic way.
15:56:57 <davdunc> not going to action it right now, but thought it was worth discussing.
15:57:03 <davdunc> #topic open floor
15:57:39 <davdunc> 4sokol this might be a great time to have you introduce yourself
15:57:55 <davdunc> very excited to have you joining us.
15:58:01 * mhayden must depart -- sorry!
15:58:06 <davdunc> bye
15:59:15 <davdunc> I think 4sokol is Matrix only, so probably not getting the ping.
15:59:38 <davdunc> alright. any other topics we need to address or haven't addressed?
16:00:08 <davdunc> anything burning a hole in your pocket or need to get off your chest?
16:00:30 <davdunc> okay. We are out of time.
16:00:35 <davdunc> #endmeeting