16:02:45 <jberkus> #startmeeting project_atomic_community_meeting
16:02:45 <zodbot> Meeting started Mon Jun 26 16:02:45 2017 UTC.  The chair is jberkus. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:45 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:02:45 <zodbot> The meeting name has been set to 'project_atomic_community_meeting'
16:02:45 <centbot> Meeting started Mon Jun 26 16:02:45 2017 UTC.  The chair is jberkus. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:45 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:02:54 <jberkus> #topic roll call
16:03:01 <jberkus> .hello jberkus
16:03:01 <zodbot> jberkus: jberkus 'Josh Berkus' <josh@agliodbs.com>
16:03:03 <yzhang> .hello yzhang
16:03:07 <zodbot> yzhang: yzhang 'Yu Qi Zhang' <jzehrarnyg@gmail.com>
16:03:08 <dustymabe> .hello dustymabe
16:03:10 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
16:03:11 <ashcrow> .hello smilner
16:03:12 <zodbot> ashcrow: smilner 'None' <smilner@redhat.com>
16:03:19 <dwalsh> .hello dwalsh
16:03:20 <zodbot> dwalsh: dwalsh 'Daniel J Walsh' <dwalsh@redhat.com>
16:03:21 <dustymabe> interesting that #topic didn't get picked up
16:03:22 <gbraad_> .hello gbraad
16:03:23 <zodbot> gbraad_: gbraad 'Gerard Braad (吉拉德)' <me+fedora@gbraad.nl>
16:03:24 <jzb> .hellomynameis jzb
16:03:25 * gbraad_ just lurking
16:03:27 <zodbot> jzb: jzb 'Joe Brockmeier' <jzb@redhat.com>
16:03:38 <rangerpb> .hello rangerpb
16:03:39 <zodbot> rangerpb: Sorry, but you don't exist
16:03:40 <tsweeney> .hello tsweeney
16:03:44 <zodbot> tsweeney: Sorry, but you don't exist
16:03:50 <mbarnes> .hello mbarnes
16:03:51 <zodbot> mbarnes: mbarnes 'Matthew Barnes' <mbarnes@redhat.com>
16:03:56 <dustymabe> :) you have to use your fedora account nick
16:04:00 <jberkus> (this is the problem with using zodbot)
16:04:05 <jberkus> but, better than nothing
16:04:22 <baude> what if the nick is already used ?
16:04:32 <dustymabe> .hello baude
16:04:33 <zodbot> dustymabe: baude 'Brent Baude' <bbaude@redhat.com>
16:04:35 <yzhang> baude: your nick doesn't matter, you can .hello as anyone
16:04:54 <ashcrow> baude: yeah just use your fedora acount in the hello command
16:04:54 <rangerpb> well then i will hello as you for now on
16:05:03 <dustymabe> :)
16:05:09 <rangerpb> fwiw i have to drop at the bottom of the hour
16:05:21 <jlebon> .hello jlebon
16:05:23 <zodbot> jlebon: jlebon 'None' <jlebon@redhat.com>
16:05:28 <miabbott> .hello miabbott
16:05:29 <zodbot> miabbott: miabbott 'Micah Abbott' <miabbott@redhat.com>
16:05:58 <jberkus> additional notes at http://etherpad.osuosl.org/atomic-community-meeting
16:06:53 <dustymabe> add yourselfs to the attendee list at the top
16:07:17 <giuseppe> .hello gscrivan
16:07:19 <zodbot> giuseppe: Sorry, but you don't exist
16:08:08 <dustymabe> aweiteka: meeting?
16:08:15 <yzhang> .hello gscrivano
16:08:16 <zodbot> yzhang: gscrivano 'Giuseppe Scrivano' <gscrivan@redhat.com>
16:08:44 <jberkus> #topic Origin vs. Kubernetes on FAH/CAH
16:08:54 <dustymabe> #chair yzhang giuseppe jberkus miabbott jlebon rangerpb mbarnes tsweeney gbraad_ ashcrow jzb
16:09:03 <maxamillion> .hello maxamillion
16:09:04 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
16:09:07 <jberkus> I'm going to limit this discussion to 20 min, because we want time for other items
16:09:40 <jberkus> #chair yzhang giuseppe jberkus miabbott jlebon rangerpb mbarnes tsweeney gbraad_ ashcrow jzb maxamillion
16:09:40 <zodbot> Current chairs: ashcrow gbraad_ giuseppe jberkus jlebon jzb maxamillion mbarnes miabbott rangerpb tsweeney yzhang
16:09:40 <centbot> Current chairs: ashcrow gbraad_ giuseppe jberkus jlebon jzb maxamillion mbarnes miabbott rangerpb tsweeney yzhang
16:10:06 <jberkus> so, everyone saw my post on atomic-devel on this topic, summarizing last week's discussion?
16:10:16 <jberkus> er, last meeting's
16:10:33 <gbraad_> yes, but it would be good to post a link anyways
16:11:11 <jberkus> will do, searching for archive link
16:11:24 <jberkus> in the meantime, does anyone have additional input on this?
16:11:39 <gbraad_> the mail was called: "[atomic-devel] Kubernetes vs. Origin for upstream"
16:12:39 <jberkus> here we go: https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-June/msg00076.html
16:12:40 <miabbott> #link https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-June/msg00076.html
16:12:54 <dustymabe> jberkus: i think you gave a pretty good summary
16:13:09 <dwalsh> I was hoping we would get more feed back from the Origin people.  The question i have is, do we have anyone actively supporting the kubernetes package?
16:13:12 <dustymabe> we did get tim st clair to volunteer to help out a bit more on the kubernetes side of things
16:13:47 <dustymabe> dwalsh: i think it's mostly jchaloup building things but not really much verification there
16:13:59 <jberkus> and jbrooks
16:14:06 <dustymabe> and then jbrooks and yzhang helping out
16:14:19 <jberkus> dwalsh: a big part of what this is about is where we're going to focus our packaging efforts
16:14:39 <ashcrow> I asked jchaloup to join us if he is able
16:14:52 <jberkus> dwalsh: AFAIK, there are no "origin" people, just OpenShift team
16:14:57 <maxamillion> jberkus: when you say "packaging" do you mean RPMs or container images?
16:15:10 <jberkus> maxamillion: well, in Fedora we need both, no?
16:15:12 <dwalsh> jberkus, Right. but if people add lots of bugzillas, and know one looks at them...
16:15:19 <jberkus> maxamillion: for Atomic, we really care about container images
16:15:47 <maxamillion> jberkus: yes, I'm just curious about the scope of the statement ... and how far/wide does it span? system containers included also?
16:15:54 <dustymabe> so here is the thing - i think if we switch our focus to origin we might end up in the same problem
16:16:07 <jberkus> maxamillion: well, it would be *mainly* system containers
16:16:45 <dustymabe> right - so if we switch our focus to origin in Fedora
16:16:49 <dustymabe> don't we have the same problem
16:16:51 <dwalsh> Either way we need to make sure we have someone signed up to maintain the system container.  Get updates into it on a regular basis, especially security ones.
16:16:56 <dustymabe> maxamillion: builds the package
16:16:59 <dustymabe> a few people test it
16:17:02 <jberkus> dustymabe: on Fedora, yes
16:17:13 <dustymabe> most people are pulling origin from docker hub
16:17:17 * dustymabe thinks
16:17:24 <dustymabe> if they are running it containerized
16:17:26 <jberkus> dustymabe: frankly, if we had the people, I'd suggest that we focus on Origin for CentOS, and Kubernetes for Fedora
16:17:31 <maxamillion> dustymabe: other than 'oc cluster up' I don't think that's accurate
16:17:33 <ashcrow> dustymabe: that is the case
16:18:05 <giuseppe> the Origin images on Docker hub already have the files necessary to run as system containers
16:18:05 <maxamillion> dustymabe: I wager that most people either get Origin from the CentOS PaaS SIG or from Fedora if they are deploying to more than one node
16:18:23 <maxamillion> giuseppe: oh?
16:18:26 <maxamillion> cool
16:18:54 <dustymabe> so what sucks about this is confusion
16:19:07 <dustymabe> if we had one funnel to send people through then it would be easier
16:19:27 <dustymabe> basically the origin team is going to make sure what is in docker hub works because that's what they promote
16:19:47 <dustymabe> none of us are on the origin team so everything we do is 'playing catch up' mostly
16:19:51 <jberkus> although "works" == "works on CentOS"
16:19:59 <giuseppe> maxamillion, https://hub.docker.com/r/openshift/{origin,node,openvswitch}
16:20:13 <maxamillion> giuseppe: awesome, thanks
16:20:25 <dustymabe> same problem with kubernetes
16:20:47 <dustymabe> which is ok, but it's going to be overhead and we need people to maintain it
16:21:10 <dustymabe> right now I personally haven't been looking at origin in fedora much at all (other than using it as a client)
16:21:30 <dustymabe> i basically set up the openshift-ansible installer from upstream and then it installs origin for me (containerized)
16:21:46 <dustymabe> which means I don't test the origin from fedora
16:21:59 <ashcrow> dustymabe: that's what I do as well
16:22:09 <dustymabe> i'm just identifying this as a potential problem
16:22:16 <jberkus> dustymabe: that's what I do,but it still frequently breaks for me
16:22:27 <dustymabe> which will end us up back in the same situation we are trying to solve now - which is reduce overhead
16:22:53 <jberkus> or expand support
16:23:22 <jberkus> Tim St. Clair weighing in points out another advantage to pursuing Kubernetes ... we can potentially get help/support from the kubernetes community
16:23:28 <jberkus> which is not true of Origin
16:23:31 <giuseppe> dustymabe, yes, and installing the system containers version should happen in the same way (openshift_use_system_containers=True to the inventory file)
16:23:53 <dustymabe> jberkus: sure
16:25:11 <dustymabe> yeah would be nice to basically build our k8s community
16:25:22 <dustymabe> if we can get more people willing to step up, then i'm 100% for keeping that around
16:25:51 <jberkus> so we're still stuck in the situation that we have two esentially equal options, but don't have the resources to pursue both
16:26:14 <jberkus> and pursuing either option adequately requires more work we're not currently doing
16:26:16 <dustymabe> jberkus: i would even venture further to say we have 3 options
16:26:30 <jberkus> what's #3?
16:26:56 <dustymabe> well, just that focusing on origin is only part of the solution since there is also fragmentation about where you can get containers from
16:27:07 <dustymabe> i.e. from docker hub or from fedora
16:27:15 <maxamillion> dustymabe: isn't there going to be fragmentation either way?
16:27:48 <dustymabe> maxamillion: not unless we just told people to pull from the origin community directly (i'm not saying that's what we should do, just pointing it as an option)
16:28:16 <jberkus> btw, I want to point out the other major cost to a focus on Origin
16:28:29 <maxamillion> dustymabe: if we want the origin stuff in registry.fedoraproject.org, I can do that this afternoon and it will go out with tomorrow's container release
16:28:32 <jberkus> right now, neither FLIBS nor CP produce Openshift-ready container images
16:28:42 <maxamillion> CP?
16:28:48 <jberkus> container pipeline.  CentOS
16:28:50 <maxamillion> oh
16:28:51 <maxamillion> right
16:29:10 <dustymabe> maxamillion: what i'm saying is that we want to get as many people as possible pulling from the same sources
16:29:19 <dustymabe> adding origin to FLIBS doesn't help us with that
16:29:21 <dustymabe> IMHO
16:29:42 <dustymabe> in order to do that we would either need to tell our users to pull from origin docker hub
16:29:45 <dwalsh> I agree, getting all of the content from Fedora is best solution.
16:29:48 <maxamillion> dustymabe: then I don't understand the problem you're stating
16:29:56 <dustymabe> or convince the origin community to tell people to pull from registry.fedoraproject.org
16:30:27 <dustymabe> i guess i'm thinking too big
16:30:31 <jberkus> or from the CentOS registry
16:30:51 <maxamillion> dustymabe: wait, are you saying that this wouldn't be a problem with k8s?
16:31:01 <jberkus> maxamillion: it would be
16:31:12 <maxamillion> jberkus: I agree
16:31:20 <dustymabe> maxamillion: yes, it would be. what i'm trying to point out is that the same problem we are trying to solve by getting rid of k8s, won't really be solved
16:31:24 <jberkus> although ... for Atomic, we can configure docker/skopeo to pull from Fedora/CentOS first
16:31:24 <maxamillion> jberkus: but I though dustymabe was saying this is an unique problem to Origin and I don't follow
16:31:43 <ashcrow> tbh I'm feeling more muddled about what we are trying to fix as the conversation goes on.
16:31:45 <jberkus> ok, it sounds like we still don't have a conclusion
16:31:56 * dustymabe will stop talking now
16:31:59 <jberkus> does anyone else have any commentary before we take this back to the mailing list?
16:32:00 <dustymabe> sorry for that
16:32:01 <jberkus> 2 minutes
16:32:04 <maxamillion> ashcrow: +1 ... I don't really understand what we're trying to solve here
16:32:24 <dustymabe> basically trying to get more alignment and spin our wheels less
16:32:25 <ashcrow> jberkus: can you reiterate the reason and the options we are aware of?
16:32:31 <maxamillion> dustymabe: ah
16:32:33 <ashcrow> jberkus: I think that will help before we go back to mailing list
16:32:54 <dustymabe> ashcrow: i can, if he doesn't want to
16:32:58 <jberkus> Currently we kind of promote Kubernetes as our main orchestrator for
16:32:58 <jberkus> upstream Atomic (Fedora Atomic and Centos Atomic).  However, we also
16:32:58 <jberkus> promote upstream Atomic as a plaform for Origin.  There are two problems
16:32:58 <jberkus> with our current approach:
16:32:59 <jberkus> 1. we don't realistically have the resources to package, document, and
16:32:59 <jberkus> promote both upstream Kubernetes and Origin on upstream Atomic.
16:33:00 <jberkus> 2. not having a default option with clear & complete documentation
16:33:00 <jberkus> confuses users and doesn't present them with a complete system.
16:33:01 <jberkus> As a result, we need to decide which platform is the one we promote
16:33:02 <jberkus> first, and ensure has a complete and up-to-date toolchain.
16:33:28 <jberkus> per email
16:33:42 <jzb> silly question
16:33:54 <jzb> do the k8s binaries from upstream "just work" on Fedora?
16:34:05 <jzb> e.g. does re-packaging it for Fedora add any value?
16:34:08 <maxamillion> FWIW, I don't personally have any interest in vanilla k8s since I only work with Origin ... I don't think that should have any impact on anyone's opinion or decision, but I won't be signing up to be doing work if things go the k8s direction and I wanted to point that out because suspect others will have a similar stance for one vs the other
16:34:32 <jberkus> jzb: no, they don't
16:34:37 <jzb> OK
16:34:43 <jberkus> jzb: hence the effort
16:34:57 <jberkus> maxamillion: that's a good question.  how many people are in maxamillion's boat?
16:35:08 <gbraad_> +1
16:35:19 <jberkus> gbraad_: what are you +1ing?
16:35:21 <ashcrow> And our options is to focus more on one at the expense of the other (for the time being). Correct?
16:35:32 <maxamillion> jzb: if they did though, repackaging adds the same value as repackaging anything in Fedora ... you get source->binary audiability and hopefully a well tested cohesive component of the distro
16:35:34 <gbraad_> I am in the same boat
16:35:38 <jberkus> ashcrow: correct.  also to promote one as the "primary" option
16:36:37 <dustymabe> i'll attempt to make them both work - but admit that i lean towards openshift
16:36:40 <ashcrow> I am in the same boat as maxamillion and gbraad_ as well.
16:37:13 <dustymabe> strigazi: around?
16:37:26 <dustymabe> we're having a conversation about kubernetes/openshift origin in project atomic
16:37:37 <maxamillion> I'm not anti-k8s by any means, it's just a lack of spare cycles and I'm already focused on openshift because I need the build pipeline stuff
16:37:52 * ashcrow nods
16:38:10 <ashcrow> I don't think anyone is anti-k8s. It's just a matter of focus and time.
16:38:24 * dustymabe thinks we need to have a kubernetes SIG or something
16:38:39 <gbraad_> maxamillion: understand. but I also do not see how it would distance people from k8s. is the involvement beneficial and helping us?
16:38:52 <jberkus> ok, we need to move on just because people mentioned other topics
16:39:04 <maxamillion> gbraad_: oh, absolutely ... it's a tough cross-roads to be at
16:39:04 <jzb> jberkus: vote or just leave open?
16:39:22 <jberkus> jzb: leave open.  folks, please continute discussion here/on the ML after the meeting
16:39:29 <jberkus> #topic upcoming events
16:39:33 <tsweeney> And which does our customer base prefer/use more?
16:39:42 <tsweeney> ooops, too late
16:40:06 <jberkus> so, we have a few events with CfPs open which we need Atomic folks for
16:40:19 <jberkus> (1) KubeCon NA (Austin, December)
16:40:33 <jberkus> (2) Container.Camp UK (London, October)
16:40:53 <maxamillion> jberkus: KubeCon NA is just down the road from me so I'll be submitting something ... just not sure what yet
16:40:56 <jberkus> I *really* want to see proposals for the various Kubernetes-involved projects for KubeCon
16:41:13 <dustymabe> cdrage: ^^ anything planned
16:41:30 <maxamillion> jberkus: it's mostly going to be a matter of if they're open to OpenShift-related talks :)
16:41:53 <jberkus> I have an open door for reviewing talk proposals
16:41:53 <maxamillion> because while OpenShift is technically a Kubernetes distribution, it's not just straight up kube
16:41:59 <maxamillion> jberkus: +1 thanks
16:42:06 <jberkus> as in, PLEASE send me your abstracts to review before you submit them
16:42:18 <jberkus> maxamillion: they've accepted OS talks before
16:42:27 <maxamillion> jberkus: cool cool
16:42:40 <jberkus> for ContainerCamp, do we have anyone in Western Europe?
16:43:07 <maxamillion> jberkus: is there any coordination with the OpenShift Team around what talks are being submitted?
16:43:32 <jberkus> maxamillion: nothing I've been able to achieve.  if you can make coordination happen, that would be great
16:43:55 <jberkus> note there will be an OpenShift Commons at Kubecon as well
16:44:03 <jberkus> so you can *also* submit talks there
16:44:10 <maxamillion> jberkus: I'll reach out and CC you
16:44:12 <jberkus> although that event focuses on OpenShift customer stories
16:44:34 <jberkus> more on mailing list
16:44:38 <jberkus> ok
16:44:43 <jberkus> #topic open florr
16:44:51 <jberkus> #topic open floor
16:45:11 <dustymabe> jberkus: was going to ask about future of atomic registry
16:45:11 <jberkus> so, a couple people wanted to raise stuff?  ashcrow, anything from commissaire?
16:45:17 <dustymabe> i don't see aweiteka around, though
16:45:30 <jberkus> dustymabe: termination, AFAIK.  we should get that formally on the agenda for next time
16:45:37 <dustymabe> yeah
16:46:01 <ashcrow> jberkus: Just that our community meeting for Commissiare is tomorrow (http://commissaire.readthedocs.io/en/latest/community_meetings.html)
16:46:03 <jberkus> #action dustymabe to get aweiteka to next meeting
16:46:07 <dustymabe> yzhang: jberkus: want to give a short summary of the linuxcon trip?
16:46:18 <gbraad_> ashcrow: I spoke with Brian Stinson at the containercon in beijing. And we wanted to be able to offer a minishift on CentOS CI. we would like to see if we can you guys on their
16:46:33 <gbraad_> s/their/there
16:46:38 <yzhang> Hm, I have the written summary, perhaps gbraad_ would have more to add
16:46:45 <yzhang> since he was also very active at the event
16:46:47 <maxamillion> I've heard that atomic registry got consumed by OpenShift and you can now deploy it as "OpenShift Stand-Alone Registry" isntead of a separate project/product --> https://docs.openshift.org/latest/install_config/install/stand_alone_registry.html
16:46:48 <dustymabe> yzhang: is that public?
16:46:55 <ashcrow> gbraad_: cool! :-)
16:46:57 <yzhang> dustymabe: no
16:46:59 <jberkus> maxamillion: that's correct
16:47:15 <dustymabe> yzhang: yeah would be good to get a public facing email maybe?
16:47:34 <dustymabe> no secret bits, just overview - anything community relevant
16:47:40 <gbraad_> dustymabe: a lot of attention for openshift and complaints about the ansible scripts
16:47:43 <maxamillion> however, last time I attempted to deploy it ... the install instructions don't work, I filed a BZ
16:47:53 <yzhang> I would say interest in openshift is far higher than atomic
16:48:03 <gbraad_> i wanted to write a summary, but didn't have the time yet
16:48:04 <yzhang> perhaps due to the fact that many people were openshift partners
16:48:07 <dustymabe> yzhang: yeah, that is to be expected
16:48:25 <yzhang> I can send out an email if you'd like
16:48:29 <dustymabe> gbraad_: what do you meanby 'offer a minishift on CentOS CI' ?
16:48:50 <gbraad_> dustymabe: a pre-installed package and running instance
16:49:04 <dustymabe> gbraad_: so for development purposes?
16:49:16 <dustymabe> or purely for test?
16:49:24 <gbraad_> this means that an install of openshift is available to target (for contained tests)
16:49:53 <gbraad_> an idea would be to have the commissaire containers run on this. but we are at the initial (discussion) stage for this.
16:50:20 <gbraad_> but according to brian, we had received similar questions from other projects
16:50:21 <thomasmckay> dustymabe: jberkus: maxamillion: i'm going to be helping out w/ issues with the registry, fyi. just getting up to speed over the next couple weeks. please add me to any discussion threads/meetings/etc.
16:50:33 <maxamillion> thomasmckay: +1 thanks
16:50:43 <dustymabe> the registry ==?
16:50:49 <thomasmckay> i don't think it is "atomic registry" specific but more "the registry ui in cockpit"
16:51:07 <gbraad_> dustymabe: will pickup this discussion with ashcrow and brian. might discuss more on the commissaire meeting
16:51:23 <jberkus> gbraad_: please also report to atomic-devel ML so we can track
16:51:35 <gbraad_> sure
16:51:45 <dustymabe> thomasmckay: ok
16:51:46 <thomasmckay> i think aweiteka was in favor or removing the "atomic registry" installation flavor and focusing on openshift usage
16:51:55 <dustymabe> yes
16:52:01 <thomasmckay> i don't know the results of those conversations, though
16:52:11 * gbraad_ will drop off now. thanks all
16:52:18 <maxamillion> thomasmckay: is that the same UI elements brought into OpenShift's Stand-Alone registry? (i.e. - does it deploy cockpit internally somehow?)
16:52:38 <dustymabe> thanks gbraad_
16:53:56 <jberkus> ok, anything else for this meeting?
16:54:05 <dustymabe> jberkus: +1 for closing
16:54:58 <tsweeney> Are we meeting next week given it's 7/3?
16:55:18 <jberkus> tsweeney: every 2 weeks
16:55:47 <jberkus> #nextmeeting is July 10th.  Topic: Atomic Registry
16:55:51 <tsweeney> Doh!, thought it was weekly, thanks!
16:55:54 <jberkus> #endmeeting