16:35:08 <dustymabe> #startmeeting fedora_atomic_wg
16:35:08 <zodbot> Meeting started Wed Jan 10 16:35:08 2018 UTC.  The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:35:08 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:35:08 <zodbot> The meeting name has been set to 'fedora_atomic_wg'
16:35:11 <dustymabe> #topic roll call
16:35:14 <jbrooks> .hello jasonbrooks
16:35:19 <zodbot> jbrooks: jasonbrooks 'Jason Brooks' <jbrooks@redhat.com>
16:35:24 <sayan> .hello sayanchowdhury
16:35:25 <zodbot> sayan: sayanchowdhury 'Sayan Chowdhury' <sayan.chowdhury2012@gmail.com>
16:35:28 <ksinny> .hello sinnykumari
16:35:29 <zodbot> ksinny: sinnykumari 'Sinny Kumari' <ksinny@gmail.com>
16:35:50 <dustymabe> .hello2
16:35:51 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
16:36:00 <dustymabe> sorry about the delay everyone. was dealing with *parking deck issues*
16:36:09 <ashcrow> .hello smilner
16:36:10 <zodbot> ashcrow: smilner 'None' <smilner@redhat.com>
16:36:20 <sanja> .hello sanja
16:36:21 <zodbot> sanja: sanja 'Sanja Bonic' <sanja@redhat.com>
16:36:35 <kushal> .hellomynameis kushal
16:36:36 <zodbot> kushal: kushal 'Kushal Das' <mail@kushaldas.in>
16:37:07 <strigazi> .hello2
16:37:08 <zodbot> strigazi: strigazi 'Spyros Trigazis' <strigazi@gmail.com>
16:37:13 <jlebon> .hello jlebon
16:37:14 <zodbot> jlebon: jlebon 'None' <jlebon@redhat.com>
16:37:26 <walters> .hello walters
16:37:26 <zodbot> walters: walters 'Colin Walters' <walters@redhat.com>
16:38:18 <dustymabe> welcome everyone :)
16:38:26 <dustymabe> #topic previous meeting action items
16:38:43 <dustymabe> from what I can tell there were no action items that were carried over from last meeting
16:39:15 <dustymabe> so we'll move forward
16:39:39 <dustymabe> #topic  Using buildah in Atomic Host
16:39:45 <dustymabe> #link https://pagure.io/atomic-wg/issue/402
16:40:10 <dustymabe> jberkus: welcome back - i think we discussed this last time when you were not here
16:40:42 <dustymabe> i think general concensus was: this seems like something that should be run in a container -
16:40:55 <dustymabe> but then also the question was raised... how big is it?
16:41:06 <dustymabe> it seems like it is 4M compressed and 12M uncompressed
16:41:12 <dustymabe> welcome unicell
16:41:28 * ashcrow nods
16:41:36 <dustymabe> #chair sanja jbrooks sayan ksinny ashcrow kushal strigazi jlebon walters unicell
16:41:36 <zodbot> Current chairs: ashcrow dustymabe jbrooks jlebon ksinny kushal sanja sayan strigazi unicell walters
16:41:50 <dustymabe> hmm did jberkus join?
16:42:12 <ashcrow> he did
16:42:23 <ashcrow> jberkus: ^^
16:42:36 <dustymabe> ashcrow: I don't see him in roll call
16:42:44 <ashcrow> dustymabe: ah, I thought you meant did he join the channel
16:43:02 <unicell> dustymabe: o/    thanks
16:43:02 <dustymabe> haha i miss the start of the meeting by a few minutes and poof
16:43:24 <ashcrow> :-)
16:43:50 <sanja> hehehe
16:44:01 <ashcrow> My $0.02 is I think going the container route makes sense, especially if we decide to move docker and cri-o to containers as well
16:44:04 <dustymabe> anyone else have any comments on this ticket?
16:44:11 <ashcrow> 12M isn't much to add to the base, but still
16:44:22 <walters> i acutally haven't played with `buildah` enough - is including just that sufficient for workflows people use with it?
16:44:24 <ashcrow> I feel like we should move to making the base image smaller
16:44:24 <sanja> do you want me to fire it up next time if you're not here? I just wanted to wait 5 polite minutes basically
16:44:44 <dustymabe> walters: i'm not sure - I assume it's just a binary so no deps??
16:44:48 <dustymabe> other than go
16:44:52 <dustymabe> well, not even that
16:44:55 <ashcrow> no deps
16:44:58 <sanja> +1 for the container route
16:45:01 <dustymabe> *static binary*
16:45:29 <walters> sorry, i meant like are there *other* tools that people commonly use with this even if the package itself doesn't actually `Requires:` them?
16:45:29 <dustymabe> sanja: sure - we usually work on in the channel who is going to run the meeting
16:45:52 <ashcrow> walters: not that come to mind.
16:45:55 <walters> e.g. some of the original usage I saw for buildah was using the host's package manager (`dnf`, `apt` etc.)
16:46:01 <dustymabe> walters: i'm guessing runc to run it once you build it?
16:46:08 <dustymabe> but probably nothing we don't already have in atomic host
16:46:10 <walters> though that doesn't seem to be the case in the current docs
16:46:18 <ashcrow> dustymabe: that's a better way of saying it :-)
16:46:35 <dustymabe> walters: that is true, that workflow might not work
16:46:42 <dustymabe> if it relies on *too much* in the host
16:46:57 <dustymabe> i imagine the most popular workflow would still be buildah bud Dockerfile
16:46:59 <ashcrow> Keep in mind that the buildah guys support running buildah in containers
16:47:08 <dustymabe> so basically do what docker does and use this dockerfile
16:47:11 <walters> yeah
16:47:47 <dustymabe> oh so if we decide to use a container, then we probably need to provide a container for people to use
16:48:07 <dustymabe> anyone want to take up that helm?
16:48:08 <walters> mmm, i dunno about that...i personally like installing rpms into a bigger tools container
16:48:39 <ashcrow> dustymabe: that's correct. While the buildah guys do support it I'm not sure if they have an official container image or not.
16:48:45 <ashcrow> walters: as in buildah + other tools?
16:48:55 <strigazi> dustymabe I can try packaging helm
16:48:55 <walters> right
16:48:57 <dustymabe> walters: i can see that - but I imagine the buildah container would require some special 'wiring' ?
16:49:13 <dustymabe> strigazi: :) helm was a figure of speech there
16:49:20 <dustymabe> wasn't referring to k8s helm
16:49:26 <dustymabe> but please do if you use it
16:49:39 <walters> true, @dwalsh did say it wanted `/var/lib/containers` mounted
16:50:17 <dustymabe> so if I create a ticket for creating a buildah container - is anyone interested in picking that work up?
16:50:33 <dustymabe> no one has to say yes here, but does make me more excited about creating the ticket if so
16:50:41 <ashcrow> dustymabe: I will do it if no one else is able to do it
16:50:51 <ashcrow> I have alot on my plate but it's important
16:50:58 <dustymabe> ashcrow: do you want to create the ticket too? and link the current topic ticket to it?
16:51:04 <ashcrow> dustymabe: sure
16:51:10 <dustymabe> ashcrow: feel free to farm it out if you can find another interested party
16:51:14 <ashcrow> dustymabe: will do
16:51:51 <dustymabe> #action ashcrow to create ticket for actually making a buildah container in fedora that we can recommend people use to build OCI images
16:52:16 <dustymabe> #topic  Overhaul list of members, Quorum Rules
16:52:25 <dustymabe> #link https://pagure.io/atomic-wg/issue/363
16:52:47 <dustymabe> we booted this one last time because jberkus wasn't around
16:53:09 <dustymabe> and it looks like jberkus gets spooked by late meetings so we are missing him this time too :(
16:53:34 <dustymabe> anyone with anything on this ticket ?
16:53:41 * dustymabe waits
16:53:54 <ashcrow> nothing really
16:54:06 <ashcrow> I think it needs jberkus unless sanja is up to speed with it already
16:54:53 <sanja> let me see
16:55:13 <sanja> let's move to next week's meeting for this
16:55:25 <dustymabe> yeah we'll move on
16:55:32 <jberkus> sorry, on conf call, readding back
16:55:59 <dustymabe> jberkus: kk
16:55:59 <sanja> and have a solution for it by then, then close the ticket
16:56:07 <jberkus> .hello jberkus
16:56:08 <zodbot> jberkus: jberkus 'Josh Berkus' <josh@agliodbs.com>
16:56:20 <jberkus> looks like the new membership quorum rules are approved
16:56:27 <jberkus> any reason not to just add them to the wiki?
16:56:59 <dustymabe> jberkus: not really - jberkus sanja? want to take an action item for that?
16:57:16 <dustymabe> keep in mind there are two proposals in that ticket
16:57:28 <dustymabe> membership proposal and voting proposal
16:57:35 <sanja> yes
16:57:37 <sanja> give it to me
16:57:45 <sanja> i'll add to the wiki then
16:58:02 <dustymabe> #action sanja to take membership proposal AND voting proposal from https://pagure.io/atomic-wg/issue/363 and add them to the wiki and close ticket
16:58:11 <dustymabe> looks good?
16:58:20 <sanja> yes, both will be added to wiki, no?
16:58:28 <sanja> and old text deleted
16:58:35 <dustymabe> sounds good
16:58:40 <dustymabe> moving on....
16:58:50 <dustymabe> #topic Decide strategy for including container runtimes in Atomic Host
16:58:55 <dustymabe> #link https://pagure.io/atomic-wg/issue/360
16:58:56 <sanja> +1
16:59:05 <dustymabe> i think we can close this one
16:59:12 <dustymabe> jberkus: agree?
16:59:18 <jberkus> ye
16:59:38 <walters> i meant to say earlier this feels related to the `buildah` discussion
16:59:39 <jberkus> I mean, we'll revisit it for F28, but not until then
16:59:56 <ashcrow> Are we adding cri-o as well to base?
17:00:28 <dustymabe> I closed ticket
17:00:50 <dustymabe> jberkus: if we were going to 'revisit it for F28' we'd need to do that now since we are in the F28 devel cycle
17:00:55 <ashcrow> I feel that the result is clear for one runtime
17:00:57 <dustymabe> ashcrow: we never decided on that
17:01:06 <ashcrow> But not for runtimes
17:01:11 <dustymabe> ashcrow: fair
17:02:04 <dustymabe> i think I would be in favor of including crio, but would really like for it to not be *HUGE*
17:02:24 <dustymabe> so depending on the size that opinion could change
17:02:40 <dustymabe> anyone have any other thoughts about this discussion?
17:02:52 <ashcrow> I recommend we make a new issue specifically about cri-o then, and rename the closed one to be specific about docker
17:03:17 <jbrooks> I agree that it'd be nice to add crio if it wasn't too huge
17:03:21 <dustymabe> +1 to new issue
17:03:38 <dustymabe> -1 to renaming old issue, since the discussion in the ticket is broader
17:03:53 <dustymabe> just give a summary comment at the bottom and link to the new crio specific issue
17:04:07 <dustymabe> WDYT?
17:04:26 <ashcrow> WFM
17:04:31 <jberkus> er, sorry, F29
17:04:36 <ashcrow> I'll create the follow on issue
17:05:01 <dustymabe> if we were to add crio to the base does that have any implications for RHELAH/CAH?
17:05:28 <unicell> just wild thought, is it possible to include things like docker in different branch?
17:05:29 <dustymabe> i.e. is there any reason we *can't* add it there
17:05:33 <jbrooks> We can do different things than them
17:05:39 <jbrooks> like removing kube
17:05:44 <dustymabe> unicell: yes it is possible
17:05:54 <dustymabe> unicell: we've been careful with that approach so far
17:06:07 <dustymabe> because it become more maintenance
17:06:12 <jberkus> mostly, it's still a lot of effort for us to publish one tree
17:06:13 <dustymabe> and starts to get more confusing
17:06:40 <unicell> I see. Basically I was thinking something similar to install non-atomic os, one can choose different category, whether it is for virtualization or desktop use
17:06:41 <dustymabe> unicell: but if our tooling/testing get a lot better then that is something we could do in the future
17:06:53 <jberkus> also more contributors!
17:06:53 <unicell> I don't see that option with Atomic unless you build your own repo
17:07:11 <dustymabe> unicell: well, we have refs in our repos now
17:07:22 <dustymabe> so we could just build more 'branches' as you say with different package sets
17:07:39 <dustymabe> we also have the yet to be implemented idea of 'server side package layering'
17:07:43 <jbrooks> That'd be cool for sure
17:07:47 <dustymabe> so a common base and then branches on top of that
17:08:03 <dustymabe> i think these are all things that we can enable when other parts of what we do get better
17:08:09 <dustymabe> mainly automated testing
17:08:13 <dustymabe> and CI
17:08:18 <unicell> yep, I understand that requires more tooling and maintanence effort. Just brought it up and wanted know whether that has been an option
17:08:19 <unicell> sure
17:08:26 <dustymabe> anywho I think we've got things worked out for this issue for now
17:08:40 <dustymabe> #action ashcrow to create issue to specifically discuss adding crio to atomic host base image
17:08:56 <dustymabe> #topic open floor / news
17:09:00 <walters> the impact on the cloud images (in particular public cloud) is a bigger ripple though
17:09:27 <dustymabe> walters: that is true, unless we just ship the base and then people add on server side branches to fit their liking
17:09:43 <dustymabe> but that all comes back to 'what is appropriate in base' :)
17:09:54 <dustymabe> I have a few items for open floor
17:09:58 <dustymabe> anyone else?
17:10:29 <walters> just one random note, i'm continuing to work on [jigdo ♲📦](https://github.com/projectatomic/rpm-ostree/issues/1081)  and it's going to unavoidably bring us to a model where we're at least downloading documentation (on the wire)
17:11:16 <walters> it's not a big deal...but it also feels weird to just discard it
17:11:23 <ashcrow> walters++
17:11:32 <walters> at least we'll be closer to being able to easily have a flag to toggle it on though it'll involve redownloading everything
17:11:34 <dustymabe> walters: good to know. I'm still interested to know when you think jigdo is something that we want to start using in Fedora
17:11:45 <walters> (this is only for FAH obviously - FAW includes docs so it just makes sense)
17:12:01 <dustymabe> for one I really really wish we could still have a 'the ostree repo is the source of truth' model for jigdo
17:12:32 <dustymabe> so jigdo would be a complement to what we have today
17:12:44 <dustymabe> however, admins don't have to use it that way if they don't
17:12:46 <dustymabe> want to
17:13:13 <dustymabe> does that make any sense?
17:13:16 <walters> once the initial experimental client side lands i think we'll be able to flesh out more of that architecture design
17:13:53 <ksinny> Is it planned to get live in F28 or later?
17:14:06 <dustymabe> ksinny: i suspect post F28
17:14:19 <ksinny> dustymabe: okay!
17:14:22 <dustymabe> most of the F28 devel stuff is already in motion
17:14:32 <walters> i know what you mean yep, but let's try to discuss it in a focused meeting on it?
17:14:32 <dustymabe> for example change proposals were due yesterday I think
17:14:47 <dustymabe> walters: indeed. I think some dedicated time at devconf would be great for this
17:14:55 <walters> with more of releng involved in particular
17:15:01 <dustymabe> jberkus: do we have a 'working session' for the atomic WG at devconf that you know of ?
17:15:09 <walters> anyways i just wanted to drop that side note
17:15:23 <dustymabe> walters: thanks!
17:15:35 <ksinny> great work walters++
17:15:35 <zodbot> ksinny: Karma for walters changed to 3 (for the f27 release cycle):  https://badges.fedoraproject.org/tags/cookie/any
17:15:48 <dustymabe> ok i have a few things for open floor
17:15:58 <dustymabe> #1 firewalld is getting closer to being in atomic host for f27
17:16:04 <dustymabe> it should be in today's rawhide build
17:16:09 <jberkus> dustymabe: nothing I submitted
17:16:40 <dustymabe> jberkus: we might find some time in the schedule and 'grab a room' if you think it would be useful, I do
17:17:12 <dustymabe> I'm going to do some tests with it in f27/rawhide to make sure I don't see any unexpected behavior and then we'll add it to the tree
17:17:26 <jberkus> dustymabe: not my decision anymore ;-)
17:17:30 <dustymabe> so firewalld may be in the next (or the next next) release of fedora atomic host
17:17:40 <dustymabe> trickling down to centos atomic host eventually
17:17:47 <dustymabe> it will be disabled by default
17:17:58 * dustymabe probably needs to make a separate blog post for this
17:17:59 <jberkus> sanja?  ^^^
17:19:01 <dustymabe> ok next thing I had
17:19:19 <dustymabe> we'll probably do another security release sometime soon for new fixes in the meltdown/spectre circus
17:19:23 <dustymabe> so look out for those
17:19:27 <sanja> yes
17:19:45 <dustymabe> and finally - anyone have any problems with any recent upgrades we have put out?
17:19:51 <dustymabe> war stories from the road?
17:20:58 <sanja> 1. not on my side and 2. regarding separate blog post yes please, disabled by default and added gotta be documented in the docs as well, so seems I gotta put that there
17:21:27 <dustymabe> sanja: :)
17:21:34 <dustymabe> i had too many streams going
17:22:00 <dustymabe> he was asking you about this: dustymabe | jberkus: we might find some time in the schedule and 'grab a room' if you think it would be useful, I do
17:22:07 <dustymabe> i.e. at devconf
17:22:40 <dustymabe> sanja: let's talk after the meeting
17:22:51 <dustymabe> ok anyone else with anything from open floor?
17:22:56 <dustymabe> ksinny: jbrooks ?
17:23:04 <dustymabe> I know there was a new CAH release out there?
17:23:05 <jbrooks> No items from me
17:23:07 * ksinny adds a note that she has started working on enabling FAH on s390x (on local machine)
17:23:12 <jbrooks> Oh, there is
17:23:39 <jberkus> wait, there still are s390s?
17:23:40 <dustymabe> ksinny: wow++
17:23:53 <ksinny> jberkus: yup
17:24:24 <ksinny> 64bit version only
17:25:14 <dustymabe> nice
17:25:26 <walters> ksinny: 👍  in a LPAR or vm?
17:25:26 <dustymabe> ok if no one has anything else we can close in a few minutes
17:25:43 <ksinny> walters: z/vm
17:26:03 <ksinny> walters: virt-install works fine on tope of z/vm, tried on F27
17:26:20 <walters> one thing related to this is AIUI historically the specialness of s390 is where regressions in e.g. dracut tend to show up
17:27:22 <ashcrow> My brain read that as :+1: in a LARP or vm
17:27:41 <dustymabe> ok closing out :)
17:27:44 <dustymabe> #endmeeting