fedora_coreos_meeting
LOGS
16:29:24 <travier> #startmeeting fedora_coreos_meeting
16:29:24 <zodbot> Meeting started Wed Apr 20 16:29:24 2022 UTC.
16:29:24 <zodbot> This meeting is logged and archived in a public location.
16:29:24 <zodbot> The chair is travier. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
16:29:24 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:29:24 <zodbot> The meeting name has been set to 'fedora_coreos_meeting'
16:29:30 <travier> #topic roll call
16:30:38 <bgilbert> .hi
16:30:39 <zodbot> bgilbert: bgilbert 'Benjamin Gilbert' <bgilbert@backtick.net>
16:30:41 <jbrooks> .hello jasonbrooks
16:30:42 <zodbot> jbrooks: jasonbrooks 'Jason Brooks' <jbrooks@redhat.com>
16:30:43 <jaimelm_> ./hello2
16:30:46 <dustymabe> .hi
16:30:47 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dusty@dustymabe.com>
16:30:56 <lorbus> .hi
16:30:56 <zodbot> lorbus: lorbus 'Christian Glombek' <cglombek@redhat.com>
16:30:59 <jaimelm> ./hello2
16:31:00 <jlebon> .hello2
16:31:03 <jaimelm> .hello2
16:31:03 <zodbot> jlebon: jlebon 'None' <jonathan@jlebon.com>
16:31:06 <zodbot> jaimelm: jaimelm 'Jaime Magiera' <jaimelm@umich.edu>
16:31:42 <saqali> .hi
16:31:43 <davdunc> .hello2
16:31:43 <zodbot> saqali: saqali 'Saqib Ali' <saqali@redhat.com>
16:31:46 <zodbot> davdunc: davdunc 'David Duncan' <davdunc@amazon.com>
16:31:48 <jlebon> #chair lorbus jaimelm saqali davdunc
16:32:14 <travier> .hi siosm
16:32:15 <zodbot> travier: Sorry, but user 'travier' does not exist
16:32:22 <travier> .hello siosm
16:32:23 <zodbot> travier: siosm 'Timothée Ravier' <travier@redhat.com>
16:32:46 <jlebon> travier: can you #chair me?
16:32:53 <travier> #chair dustymabe lorbus bgilbert jbrooks jaimelm jlebon
16:32:53 <zodbot> Current chairs: bgilbert dustymabe jaimelm jbrooks jlebon lorbus travier
16:33:09 <travier> #chair saqali davdunc
16:33:09 <zodbot> Current chairs: bgilbert davdunc dustymabe jaimelm jbrooks jlebon lorbus saqali travier
16:33:10 <jlebon> #chair saqali davdunc
16:33:10 <zodbot> Current chairs: bgilbert davdunc dustymabe jaimelm jbrooks jlebon lorbus saqali travier
16:33:15 <jlebon> heh
16:33:59 <jlebon> let's give it another 45s
16:35:25 <jlebon> #topic Action items from last meeting
16:35:30 <jlebon> -ENOENT
16:35:37 <jlebon> great!
16:35:52 <gursewak> .hi
16:35:53 <zodbot> gursewak: gursewak 'Gursewak Singh' <gurssing@redhat.com>
16:36:05 <jlebon> let's move on to ticket issues
16:36:12 <jlebon> #topic tracker: Fedora 36 changes considerations
16:36:14 <jlebon> #link
16:36:17 <jlebon> #undo
16:36:17 <zodbot> Removing item from minutes: <MeetBot.items.Link object at 0x7fccadd892e8>
16:36:20 <jlebon> #link https://github.com/coreos/fedora-coreos-tracker/issues/918
16:36:28 <jlebon> dustymabe: want to take this one?
16:37:37 <dustymabe> jlebon: actually I should probably remove that one from meeting agenda
16:37:47 <dustymabe> podman 4.0 - no complaints so far
16:37:52 <dustymabe> DNS over TLS got moved to F37
16:38:07 <dustymabe> and we'll get updates about the ppc change from ravanelli as they come
16:38:15 <aaradhak> .hi
16:38:16 <zodbot> aaradhak: aaradhak 'Aashish Radhakrishnan' <aaradhak@redhat.com>
16:38:21 <dustymabe> i'll drop the meeting label
16:38:38 <jlebon> dustymabe: ack thanks
16:38:44 <jlebon> let's move on then
16:38:52 <travier> #chair aaradhak gursewak
16:38:52 <zodbot> Current chairs: aaradhak bgilbert davdunc dustymabe gursewak jaimelm jbrooks jlebon lorbus saqali travier
16:38:53 <jlebon> #topic New Package Request: nmstate-libs and nmstate
16:38:56 <jlebon> #link https://github.com/coreos/fedora-coreos-tracker/issues/1175
16:39:16 <jlebon> does someone want to introduce this?
16:39:45 <travier> (I should have asked the team to join this meeting)
16:39:59 <travier> I'll make a short intro but I don't have the full context
16:40:29 <travier> The nmstate team is asking us to include nmstate as part of FCOS
16:40:44 <travier> nmstate is a project to declaratively configure the network
16:41:34 <travier> Having that as part of FCOS would enable easier network configuration in a fashion that would be similar to what systemd-networkd offered
16:42:07 <travier> Previously this was written in Python so the discussion was over. Now this is Rust thus this can be considered again
16:42:14 <travier> EOI
16:42:19 <travier> (End of intro)
16:42:30 <jlebon> travier: thanks
16:42:33 <bgilbert> as I mentioned in the ticket, we'll need to discuss how nmstate would integrate with the OS
16:42:50 <bgilbert> the proposal claims that no services would be shipped, but that's not exactly useful when provisioning via Ignition
16:43:25 <bgilbert> I'm hoping there's some service we can ship (maybe enabled by default, maybe not), so that Ignition can write an nmstate config and it'll be auto-applied
16:43:36 <dustymabe> agree ^^
16:43:37 <jlebon> i should really know this but: does nmstate just produce NM keyfiles underneath, or does it directly drive NM through its APIs?
16:43:45 <travier> bgilbert: +1
16:43:50 <dustymabe> jlebon: I don't know the answer to that question either
16:43:57 <bgilbert> jlebon: I seem to recall it's the latter, but don't quote me
16:44:19 <dustymabe> i'd also like to know how nmstate config intersects with NM keyfiles (if they both exist) and what not
16:44:25 <bgilbert> +1
16:44:28 <jlebon> bgilbert: or maybe better, have nmstate ship that service
16:44:29 <dustymabe> travier: let's try to get the team into our next meeting?
16:44:46 <jlebon> since it'd be equally relevant in other provisioning systems
16:44:47 <travier> dustymabe: +1
16:45:16 <jlebon> bgilbert: i think that's the same thing you said, so disregard :)
16:45:47 <jlebon> so should we revisit this in a future meeting with the right folks attending?
16:46:10 <dustymabe> +1 from my side - can we action someone to followup with the team and make sure they can make it?
16:46:53 <jlebon> travier: would you like to take that?
16:47:38 <travier> I'll be out for 2 weeks so someone else should probably work on that
16:48:30 <jlebon> ok I'll take it
16:48:32 <dustymabe> I can.. but also if someone new wants to see how things are done.. feel free to volunteer and I'll show you the way :)
16:48:41 <dustymabe> or jlebon can show you the way :)
16:48:48 <jlebon> :)
16:49:09 <jlebon> #action jlebon to reach out to nmstate team to make sure they can attend a future meeting to discuss this ticket
16:49:13 <jlebon> ok let's move on
16:49:24 <jlebon> #topic  develop strategy around organization and naming for our containers in quay.io
16:49:27 <jlebon> #link https://github.com/coreos/fedora-coreos-tracker/issues/1171
16:49:33 <jlebon> dustymabe: want to introduce that one?
16:50:40 <dustymabe> yeah, basically let's figure out what all containers we have, what containers we want to have in the future, what organization we want them under, and what name we want to give them
16:50:59 <dustymabe> there's a few topics to unpack here
16:51:20 <dustymabe> 1. where do we thing coreos-assembler container should be long term? do we need to keep the coreos-assembler org?
16:51:38 <dustymabe> 2. where do we think the new kubevirt container should go? what should we name it so it's clear what it is?
16:52:06 <dustymabe> 3. where do we want the ostree container of FCOS (the basis for CoreOS Layering) to go? what should we name it so it's clear what it is?
16:52:51 <dustymabe> also 4. Is Fedora going to move containers from registry.fedoraproject.org to quay.io/fedora eventually?
16:53:14 <dustymabe> EOF
16:53:40 <jlebon> i think 2 and 3 really should be in the same registry
16:53:59 <dustymabe> jlebon: same registry or same org?
16:54:39 <jlebon> having cosa under coreos/ would be nice, but it's been useful for our team to have admin access to the org, which I'm not sure we'd have with coreos/
16:55:00 <jlebon> and i'm not sure if it's worth the churn. we'd probably have to keep mirror it there for a long time
16:55:03 <dustymabe> does anyone know the answer to 4. or have any more information about it? walters?
16:55:31 <jlebon> dustymabe: same registry and org, yeah
16:55:35 <dustymabe> jlebon: ok
16:56:23 <walters> is it really that it's just bgilbert has admin access to coreos/ ?
16:56:36 * bgilbert doesn't have it either
16:56:51 <walters> btw this container thing also includes our containers for butane and coreos-installer I think
16:57:15 <dustymabe> walters: those are listed in the ticket (under coreos/ org already)
16:57:27 <bgilbert> walters: eh, maybe.  the upstream components are arguably separate
16:57:35 <jlebon> i think those makes sense to keep under coreos/
16:57:42 <bgilbert> butane, coreos-installer, ignition-validate
16:57:53 <bgilbert> and mkpasswd I suppose
16:58:31 <dustymabe> ok so i'm seeing a theme - let me see if I can pull away some conclusions and re-focus the discussion
16:58:54 <dustymabe> for 1. we're fine leaving coreos-assembler where it is for now and figuring out if we want to move it later
16:59:07 <dustymabe> we don't yet know the answer to 4.
16:59:25 <dustymabe> for 2. and 3. we think they should go in the same registry/org (live in the same place)
16:59:42 <dustymabe> now the question is what registry/org and what to name them
17:00:17 <dustymabe> should we just say we'll put them wherever fedora puts its containers (if it stays reg.fp.o then we put them there, if it moves to quay, we move with the rest?)
17:00:49 <dustymabe> retyping for bgilbert
17:00:51 <dustymabe> should we just say we'll put them wherever fedora puts its containers (if it stays reg.fp.o then we put them there, if it moves to quay, we move with the rest?)
17:01:02 <bgilbert> +1
17:01:57 <jlebon> +1
17:02:27 <dustymabe> i'm not sure what level of control or flexibility we'll have by doing that, but we can at least find out and also find out what the strategy for 4. is
17:02:43 <dustymabe> any opposed?
17:02:51 <bgilbert> I think there's value for ?branding? reasons, even if we lose some control
17:03:15 <dustymabe> +1
17:03:22 <dustymabe> ok now - naming
17:03:33 <dustymabe> one is for kubevirt (any other possible uses?)
17:03:47 <dustymabe> and one is for layering/other utilities?
17:05:12 <dustymabe> registry.example.com/fedora-coreos-containerdisk
17:05:14 <dustymabe> registry.example.com/fedora-coreos-ostree
17:05:16 <dustymabe> ?
17:05:21 <dustymabe> oops let me fix that up
17:05:51 <dustymabe> well i guess it depends on the registry - if it's reg.fp.o it would look something like ^^
17:05:57 <dustymabe> if quay then something like:
17:06:06 <jlebon> i think i'd want 'kubevirt' in the name -- `fedora-coreos-kubevirt`?
17:06:08 <bgilbert> s/containerdisk/kubevirt/.  we name images by the platform, not the file format.
17:06:10 <bgilbert> +1
17:06:25 <dustymabe> quay.io/fedora/fedora-coreos-containerdisk
17:06:27 <dustymabe> quay.io/fedora/fedora-coreos-ostree
17:06:45 <dustymabe> bgilbert: good point
17:06:51 <jlebon> not sure we need the -ostree prefix
17:07:23 <dustymabe> IOW if there was a new platform that used a containerdisk format and we pushed to the registry we'd need a separate containerdisk (platform might have specific behavior)
17:07:36 <jlebon> it's a regular container as is that you can play with
17:07:41 <bgilbert> dustymabe: even if it didn't have specific behavior, we'd burn in a separate platform ID
17:07:56 <bgilbert> into the disk image
17:08:01 <bgilbert> kargs
17:08:17 <dustymabe> bgilbert: +1
17:08:46 <dustymabe> so fedora-coreos-kubevirt - and we shouldn't need to make that arch specific because we can use manifest lists
17:08:57 <bgilbert> yup
17:09:38 <dustymabe> ok and for the other one
17:09:51 <dustymabe> `fedora-coreos-ostree` - not sure if we need the `ostree` prefix
17:10:06 <dustymabe> I think I agreed originally - but now am not so sure
17:10:43 <dustymabe> i guess with `fedora-coreos-kubevirt` being specific then one probably wouldn't confuse that with `fedora-coreos`
17:10:51 <dustymabe> so I'm ok with that if that's what we want
17:11:42 <dustymabe> ok so here's what we would probably end up with
17:11:45 <dustymabe> either:
17:11:51 <dustymabe> registry.fedoraproject.org/fedora-coreos-kubevirt
17:11:53 <dustymabe> registry.fedoraproject.org/fedora-coreos
17:11:55 <dustymabe> or:
17:11:58 <dustymabe> quay.io/fedora/fedora-coreos-kubevirt
17:12:00 <dustymabe> quay.io/fedora/fedora-coreos-ostree
17:12:02 <dustymabe> oops
17:12:09 <dustymabe> quay.io/fedora/fedora-coreos
17:12:37 <dustymabe> does that match the discussion?
17:12:54 <jlebon> +1 that sounds good to me!
17:13:25 <jaimelm> +1
17:14:06 <jlebon> re. -ostree, i can see that if it were the format used by RHCOS currently (archive repo in a container). i.e. it's just a fancy wrapper around an ostree commit, but the new container format is a valid container in its own right
17:14:25 <bgilbert> dustymabe: +1
17:15:05 <dustymabe> #proposed We have decided that we'll punt on long term strategy for coreos-assembler container location. We'll reach out to Fedora infra/releng to investigate long term strategy for what registry Fedora containers will be hosted in. For the kubevirt and ostree containers we'll host them wherever Fedora officially hosts its containers and name them `fedora-coreos-kubevirt` and
17:15:07 <dustymabe> `fedora-coreos`.
17:15:49 <jlebon> ack
17:16:28 <dustymabe> any nays ?
17:17:03 <bgilbert> ack
17:17:08 <dustymabe> #agreed We have decided that we'll punt on long term strategy for coreos-assembler container location. We'll reach out to Fedora infra/releng to investigate long term strategy for what registry Fedora containers will be hosted in. For the kubevirt and ostree containers we'll host them wherever Fedora officially hosts its containers and name them `fedora-coreos-kubevirt` and
17:17:10 <dustymabe> `fedora-coreos`.
17:17:30 <jlebon> cool, thanks for driving this dustymabe!
17:17:38 <jlebon> let's move on
17:18:10 <jlebon> dustymabe: did you want to talk about https://github.com/coreos/fedora-coreos-tracker/issues/1154 again or should we untag it?
17:18:19 <jlebon> ahh we're past CfP anyway :)
17:18:53 <jlebon> that takes us to
17:18:55 <jlebon> #topic Open Floor
17:19:05 <dustymabe> jlebon: actually the CFP got extended
17:19:12 <walters> (sorry, got pulled away) I just do want to strongly agree in dropping the `-ostree` suffix; a big part of the goal of this effort is to move ostree more into the background and not be something everyone needs to understand.  The goal is you boot a container, you upgrade to a new container - there's some ostree stuff in the background.
17:19:44 <jlebon> dustymabe: we can chat about it in open floor I guess :)
17:20:03 <jlebon> walters: +1
17:20:05 <dustymabe> jlebon: agreed
17:20:17 <dustymabe> i submitted a generic FCOS intro/current state talk
17:20:22 <dustymabe> skunkerk submitted a lab session
17:20:26 <jlebon> looks like CfP deadline is today
17:20:39 <dustymabe> would be interesting to submit a talk on coreos-layering
17:20:43 <jlebon> nice dustymabe++ skunkerk++
17:21:56 <jlebon> i likely won't be able to attend those dates, so didn't submit anything
17:22:00 <dustymabe> I guess we could try to find someone to pick up the torch on the container registry discussion we just had and interface with fedora infra/releng on their plans
17:22:49 <dustymabe> anybody interested in learning some inner workings of Fedora and meeting the releng/infra teams?
17:23:02 * jaimelm throws in his hat
17:23:12 <jaimelm> I'm back to be able to attend meetings.
17:23:23 <jaimelm> jettisoned other responsibliities.
17:23:27 <dustymabe> also ICYMI: this was pretty cool: minishift automation on top of FCOS: https://discussion.fedoraproject.org/t/automated-fedora-coreos-35-for-raspberry-pi-4-b-400/38359/4
17:23:35 <dustymabe> jaimelm: sweet
17:24:08 <dustymabe> #action jaimelm and dustymabe to meet with the releng/infra team to talk about containers and where everything fits
17:24:20 * jaimelm thumbs up
17:24:25 <jlebon> +1 thanks!
17:24:39 <dustymabe> the guy from that minishift post is willing come give a demo at one of our meetings
17:25:07 <dustymabe> anybody want to volunteer to organize a video meeting (come up with an agenda) for May 4th?
17:25:14 <jaimelm> Is there a video meeting May 4th?
17:25:18 <jaimelm> ahh
17:25:36 <dustymabe> jaimelm: we try to have them first wednesday of the month, but there isn'
17:25:40 <jlebon> dustymabe: super cool
17:25:45 <dustymabe> isn't always someone who can organize it
17:25:49 <jaimelm> right
17:26:42 <jlebon> maybe aaradhak would be interested?
17:27:36 <jlebon> or gursewak?
17:27:47 <aaradhak[m]> yes I can.
17:28:11 <gursewak> I can take next one then:)
17:28:51 <dustymabe> +1 - you can both learn the steps this time.. aaradhak[m] can you set up some time later this week for us to organize it?
17:28:58 <dustymabe> invite gursewak and myself
17:29:33 <aaradhak[m]> sure will set up a meeting
17:29:40 <jlebon> cool, thanks all! :)
17:29:49 <jlebon> anything else before I close the meeting?
17:30:03 <jlebon> (closing in 30s otherwise)
17:30:33 <jlebon> #endmeeting