16:32:02 <dustymabe> #startmeeting fedora_coreos_meeting
16:32:02 <zodbot> Meeting started Wed Jan  2 16:32:02 2019 UTC.
16:32:02 <zodbot> This meeting is logged and archived in a public location.
16:32:02 <zodbot> The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:32:02 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:32:02 <zodbot> The meeting name has been set to 'fedora_coreos_meeting'
16:32:10 <bgilbert[1]> .hello bgilbert
16:32:11 <zodbot> bgilbert[1]: bgilbert 'Benjamin Gilbert' <bgilbert@backtick.net>
16:32:11 <dustymabe> #topic roll call
16:32:15 <dustymabe> .hello2
16:32:16 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dusty@dustymabe.com>
16:32:23 <ksinny> .hello sinnykumari
16:32:24 <zodbot> ksinny: sinnykumari 'Sinny Kumari' <ksinny@gmail.com>
16:32:34 <kaeso> .hello lucab
16:32:35 <zodbot> kaeso: lucab 'Luca Bruno' <lucab@redhat.com>
16:32:46 <miabbott> .hello2
16:32:47 <zodbot> miabbott: miabbott 'Micah Abbott' <miabbott@redhat.com>
16:33:36 <slowrie> .hello2
16:33:37 <zodbot> slowrie: slowrie 'Stephen Lowrie' <slowrie@redhat.com>
16:33:44 <dustymabe> #chair bgilbert[1] ksinny kaeso miabbott slowrie
16:33:44 <zodbot> Current chairs: bgilbert[1] dustymabe kaeso ksinny miabbott slowrie
16:35:09 * dustymabe waves at jligon
16:35:22 <jligon> .hello2
16:35:23 <zodbot> jligon: jligon 'Jeff Ligon' <jligon@redhat.com>
16:35:27 <dustymabe> #chair jligon
16:35:27 <zodbot> Current chairs: bgilbert[1] dustymabe jligon kaeso ksinny miabbott slowrie
16:35:30 * jligon waves back
16:35:49 <sayan> .hello sayanchowdhury
16:35:49 <dustymabe> #topic Action items from last meeting
16:35:49 <zodbot> sayan: sayanchowdhury 'Sayan Chowdhury' <sayan.chowdhury2012@gmail.com>
16:35:54 <dustymabe> #chair sayan
16:35:54 <zodbot> Current chairs: bgilbert[1] dustymabe jligon kaeso ksinny miabbott sayan slowrie
16:36:07 <dustymabe> here are the action items from last meeting
16:36:12 <dustymabe> #link https://meetbot-raw.fedoraproject.org/teams/fedora_coreos_meeting/fedora_coreos_meeting.2018-12-19-16.30.txt
16:36:21 <dustymabe> * jlebon to ask rpm-software-management org on github about name move of
16:36:22 <dustymabe> rpm-ostree repo
16:36:24 <dustymabe> * dustymabe to summarize GC discussion here today in #99
16:36:26 <dustymabe> * lucab to check/document why CL images are garbage-collected on gcloud
16:36:45 <dustymabe> #info dustymabe summarized GC discussion in #99
16:36:53 <kaeso> ah, I did not
16:36:56 <dustymabe> #link https://github.com/coreos/fedora-coreos-tracker/issues/99#issuecomment-448717198
16:37:06 <dustymabe> kaeso: re-action ?
16:37:10 <kaeso> bgilbert[1]: did you maybe ask crawford about that?
16:37:57 <kaeso> dustymabe: unless Benjamin did that already, yes
16:38:12 <dustymabe> +1
16:38:19 <dustymabe> #action kaeso to check/document why CL images are garbage-collected on gcloud
16:38:32 <dustymabe> i'll re-action for jlebon too since he seems to be AFK
16:38:41 <dustymabe> #action jlebon to ask rpm-software-management org on github about name move of rpm-ostree repo
16:38:46 <mnguyen_> .hello mnguyen
16:38:47 <zodbot> mnguyen_: mnguyen 'Michael Nguyen' <mnguyen@redhat.com>
16:38:51 <bgilbert[1]> kaeso: I didn't
16:38:52 <dustymabe> #chair mnguyen_
16:38:52 <zodbot> Current chairs: bgilbert[1] dustymabe jligon kaeso ksinny miabbott mnguyen_ sayan slowrie
16:39:49 <dustymabe> ok there aren't any tickets with meeting labels today (just coming back from break) but there are a few things I'd like to run through
16:39:56 <dustymabe> #topic Garbage collection policy for OS releases
16:40:00 <dustymabe> #link https://github.com/coreos/fedora-coreos-tracker/issues/99
16:40:13 <dustymabe> so we summarized the discussion from last time and requested releng review
16:40:30 <dustymabe> I didn't think we'd get anything back since mohan was already out but it looks like he came up with a brief response
16:40:40 <dustymabe> anything more we'd like to reply to him with for now?
16:41:40 <bgilbert[1]> I'm not clear on whether that comment entirely answered my questions
16:42:09 <dustymabe> bgilbert[1]: maybe you and I can follow up in ticket?
16:42:23 <bgilbert[1]> in particular, for the non-"devel" artifacts: are old update packages not pruned?
16:42:28 <dustymabe> i think in general it looks like fedora's current policy doesn't differ too much from what we want to do, right
16:42:32 <bgilbert[1]> can do, unless someone knows off the top of their head
16:42:39 <bgilbert[1]> pruned earlier, I mean
16:42:49 <bgilbert[1]> dustymabe: yeah
16:43:33 <dustymabe> bgilbert[1]: one thing that might be confusing is the use of the word 'artifact'
16:43:57 <dustymabe> so when fedora cuts a release it includes all rpms and media that were created
16:44:16 <dustymabe> but then updates happen and new rpms are put in the repos
16:44:47 <dustymabe> so you have the release day artifacts (including media + rpms) and updates yum repos (rpms)
16:45:02 <dustymabe> i think mohan is mostly referring to the policy for 'release day artifacts'
16:45:20 <dustymabe> i'm not sure what the policy is for update artifacts
16:45:38 <dustymabe> but in general I think we can work together to make sure our needs are satisfied
16:45:58 <dustymabe> #action bgilbert[1] dustymabe to follow up with releng on GC policy in #99
16:46:08 <bgilbert[1]> hmm.  so for us there are _only_ "release day" RPMs, it's just that we'll do a lot of releases :-)
16:46:12 <bgilbert[1]> dustymabe: +1
16:46:50 <dustymabe> bgilbert[1]: right, but our primary artifactt is actually not an rpm, it's an ostree. so as long as we keep the ostree + golden images then I don't think we need the rpms
16:47:06 <dustymabe> but we can discuss :)
16:47:07 <walters> .hello2
16:47:08 <zodbot> walters: walters 'Colin Walters' <walters@redhat.com>
16:47:13 <dustymabe> welcome walters
16:47:16 <dustymabe> #chair walters
16:47:16 <zodbot> Current chairs: bgilbert[1] dustymabe jligon kaeso ksinny miabbott mnguyen_ sayan slowrie walters
16:47:49 <bgilbert[1]> +1 to moving on
16:48:01 <dustymabe> #topic roadmap for FCOS
16:48:10 * dustymabe hopes this link works
16:48:16 <dustymabe> https://github.com/coreos/fedora-coreos-tracker/pull/96/files?short_path=38574c0#diff-38574c080d4e2eb38c49b86e6588ad98
16:48:36 <dustymabe> ^^ this is the current PR with the roadmap for Fedora CoreOS
16:48:50 <dustymabe> maybe I should call it a timeline and not a roadmap, but you get the idea
16:49:08 <dustymabe> this will be a WIP and will get changed often
16:49:25 <dustymabe> the important thing is that there are no "must have" items for FCOS that are not on that list
16:49:47 <dustymabe> so if there are any missing please let's get them added
16:50:12 <dustymabe> the schedule is currently front heavy, because a lot of this will
16:50:17 <dustymabe> "discover" new work
16:50:46 <bgilbert[1]> one thing jumps out atm: live PXE
16:50:51 <bgilbert[1]> for which we do not have a ticket
16:50:59 <dustymabe> +1
16:51:01 <bgilbert[1]> #action bgilbert[1] to file a ticket for live PXE support
16:51:12 <dustymabe> <3
16:51:38 <dustymabe> we're going to find gaps, as bgilbert[1] just did
16:51:52 <dustymabe> the important part is that we consistently churn through the items we do have on our plate
16:52:10 <ksinny> +1
16:52:16 <dustymabe> if there are things that aren't must haves for final release, let's make sure we identify them as things we can do after release and be OK
16:52:59 <dustymabe> we have lofty goals right now, some things will most likely drop off the plate. Let's make sure those are ones we can live without :)
16:53:20 <dustymabe> live without *temporarily*
16:53:37 <dustymabe> any comments before moving to next topic?
16:54:05 <dustymabe> actually - one thing to mention
16:54:24 <dustymabe> this timeline will probably start to dominate our meeting schedule a bit (i.e. checking in on progress and such of planned items)
16:54:35 <dustymabe> so we may have a mix of "meeting"
16:54:46 <dustymabe> label topics, + timeline related discussion
16:55:49 <dustymabe> and that's all I had for large level topics
16:56:03 <dustymabe> #topic open floor
16:56:42 <dustymabe> anyone with items for open floor - we'll go into each discussion as a separate topic to keep focus
16:58:07 <dustymabe> ok I'll bring a few up then
16:58:17 <dustymabe> minitopic: ostree mirroring
16:58:33 <dustymabe> we've had some discussion in various tickets on this topic before the break
16:58:56 <dustymabe> it seems like we're oscillating a bit between double down on CDN or attempt mirroring static delta repos
16:59:15 <dustymabe> ksinny was going to do some investigation over the past two weeks
17:00:21 <dustymabe> if we want to tweak and experiement with CDN we'll need to wait til some of fedora infra is back from holiday which should be in a few weeks
17:00:23 <ksinny> yes, CDN path is already in use fr Atomic Host hence easier but we are looking into mirrormanager which requires investigation around static delta only repos
17:00:55 <ksinny> #info ostree mirroring support in mirrormanager is being discussed in https://github.com/fedora-infra/mirrormanager2/issues/258
17:01:28 <dustymabe> cool
17:01:37 <walters> we don't technically need to wait, can use our own creds to set up testing stuff right?
17:02:19 <dustymabe> walters: yeah - that is true. we can set up a separate CDN in cloudfront and test a different configuration
17:02:36 <dustymabe> I've done it before. It's not hard to do
17:02:57 <dustymabe> ksinny: if you want to try I can help you set this up and make sure your account has the right permissions
17:03:12 <ksinny> dustymabe: yes, I was about to say same :)
17:03:59 <dustymabe> +1
17:04:00 <dustymabe> thanks ksinny
17:04:20 <ksinny> thanks dustymabe!
17:04:21 <dustymabe> i'll just keep running through small topics unless someone stops me or has another topic
17:04:33 <dustymabe> minitopic: coreos-metadata as an rpm
17:04:46 <dustymabe> #link https://github.com/coreos/fedora-coreos-tracker/issues/4
17:04:55 <kaeso> o/
17:05:02 <dustymabe> jlebon pointed out to me in the roadmap PR that I should add this to the roadmap
17:05:30 <dustymabe> so I added it, just wondering where we stand on it and who can help
17:06:06 <kaeso> the RPM I did is for RHCOS, as the Fedora policy for Rust packages is quite different
17:06:30 <kaeso> I finished updating all the ancient dependencies
17:06:48 <kaeso> so it should be a bit easier to package them in Fedora
17:06:59 <dustymabe> nice!
17:07:00 <kaeso> (I didn't check which one are still missing though)
17:07:10 <kaeso> *ones
17:07:16 <dustymabe> kaeso: actually I found some notes I wrote before the break
17:07:38 <walters> you're going with the packaged crates?
17:08:33 <dustymabe> kaeso: just commented in the public issue with those notes: https://github.com/coreos/fedora-coreos-tracker/issues/4#issuecomment-450922699
17:08:36 <kaeso> walters: from what I've read, that's the current Fedora policy
17:09:26 <walters> it's not required no
17:10:18 <kaeso> walters: my knowledge comes from https://fedoraproject.org/wiki/Packaging:Rust#Naming
17:10:47 <walters> there's some more on this in https://fedoraproject.org/wiki/Bundled_Libraries?rd=Packaging:Bundled_Libraries
17:11:07 <dustymabe> kaeso: it says 'Rust libraries MUST be from crates.io'
17:11:18 <dustymabe> but is coreos-metadata a "rust library" ?
17:11:21 <dustymabe> or just a binary
17:12:45 <kaeso> dustymabe: I think it fits into https://fedoraproject.org/wiki/Packaging:Rust#Binary
17:13:57 <dustymabe> right. so I think if nothing else is going to use this package to 'build' or by dynamically linked to then we are OK to bundle
17:14:26 <dustymabe> because we aren't going to be a 'rust library' and we don't need to be named rust-coreoso-metadata
17:14:43 <dustymabe> ^^ let me know if my interpretation is wrong
17:15:04 <dustymabe> the analogy i'm drawing here is
17:15:33 <dustymabe> docker is named docker, not golang-docker because the package is just delivering a binary
17:16:40 <kaeso> dustymabe: I think the guideline is speaking about source rpm name
17:17:15 <dustymabe> ok
17:17:33 <dustymabe> either way. I would prefer to not bundle if we could.
17:17:54 <dustymabe> it looks like we are 3 deps away so maybe we just knock those out
17:18:56 <dustymabe> kaeso: walters any other comments before we move on?
17:19:27 <kaeso> nope
17:19:46 <dustymabe> minitopic: tracker for 'cloud'/'no cloud agents' work
17:20:06 <dustymabe> I created a tracker for this work with a lot of little check boxes and such
17:20:10 <dustymabe> https://github.com/coreos/fedora-coreos-tracker/issues/95
17:20:37 <dustymabe> this should help us keep up with what has/hasn't been done
17:20:46 <bgilbert[1]> - [ ] checkboxes!
17:20:59 <dustymabe> if you find new items related to "clouds" that needs to be done, please add it or let me know about it
17:21:26 <ksinny> lots of checkbox :)
17:21:33 <bgilbert[1]> ... s/\[ /\[x/ of course
17:21:33 <zodbot> bgilbert[1]: Error: Missing "]".  You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands.
17:21:54 <dustymabe> silly zodbot ] are for kids
17:21:55 <bgilbert[1]> zodbot: kay
17:22:22 <dustymabe> ok i'll close out the meeting a bit early if no one has any other topics
17:22:23 * jbrooks just now looks at clock
17:22:29 <dustymabe> oh hai jbrooks
17:22:31 <dustymabe> :)
17:22:34 <jbrooks> :)
17:22:52 <dustymabe> jbrooks: we're in open floor
17:22:58 <dustymabe> anything you'd like to bring up?
17:23:02 <dustymabe> #chair jbrooks
17:23:02 <zodbot> Current chairs: bgilbert[1] dustymabe jbrooks jligon kaeso ksinny miabbott mnguyen_ sayan slowrie walters
17:23:27 <jbrooks> Not really, I've still been thinking / wondering about the okd/kube stuff
17:24:03 <kaeso> jbrooks: https://github.com/coreos/fedora-coreos-tracker/issues/93 ?
17:24:07 <dustymabe> jbrooks: yeah me as well
17:24:31 <jbrooks> I wonder if package layering wouldn't be so bad -- there isn't an easy option for the flexibility
17:24:52 <jbrooks> And by maintaining certain repos. we could let ppl stay on particular versions of things
17:25:11 <jbrooks> kaeso, that's the right issue
17:25:52 <jbrooks> Another thing, with OKD 4.0, we don't have packages for that yet, so we're still on the ansible side of things for cluster setup
17:26:55 <dustymabe> jbrooks: yeah package layering is one of the ways we talked about allowing people to "switch versions" of kube
17:27:31 <dustymabe> I still think it would be nice to have it be "included" in whatever artifact we deliver, though
17:27:44 <jbrooks> That still works even if we do bake one in -- you can layer to replace
17:27:46 <dustymabe> "included" could mean various things, from baked in to the ostree
17:28:06 <dustymabe> jbrooks: right, you can layer to replace
17:28:19 <dustymabe> I don't think that differs from what we have talked about in the past
17:28:30 <jbrooks> Right
17:28:41 <dustymabe> the one thing we have said that we want to make sure people can do is "pin" to a version of kubernetes/origin though
17:29:21 <dustymabe> i.e. if you are on kube 1.8 and the next FCOS has 1.9 baked in and you want to stay, then there should be some way to configure that
17:29:27 <jbrooks> That would require a layering approach, I think
17:29:53 <jbrooks> Or, baked in, and if you want to opt out of the baked in update, you'll need to switch to layered
17:30:14 <dustymabe> right
17:30:22 <dustymabe> so the devil here is in the details mostly
17:30:39 <jbrooks> Anyway, before we know it, we'll be shipping, so we need to figure this stuff out
17:31:02 <dustymabe> but i think in general this was the strategy we were thinking of taking (with modularity allowing for users to select different streams possibly)
17:31:08 <dustymabe> https://github.com/projectatomic/rpm-ostree/issues/1466
17:32:07 <dustymabe> jbrooks: i agree. the sooner we figure this out the better
17:32:31 <dustymabe> jbrooks: we could use some help in driving this conversation, thanks for your nudges in that direction so far
17:32:51 <dustymabe> jbrooks: will you help us head up this effort?
17:33:01 <jbrooks> jbrooks, Yes
17:33:05 <jbrooks> heh
17:33:09 <jbrooks> dustymabe, Yes
17:33:14 <dustymabe> :)
17:33:31 <jbrooks> still getting reacquainted w/ the keyboard
17:33:36 <dustymabe> cool - feel free to add meeting labels to items and we'll discuss early/often
17:33:47 <jbrooks> OK
17:33:58 <dustymabe> any other topics before we close out the first meeting of 2019?
17:35:01 <dustymabe> #endmeeting