fedora_atomic_wg
LOGS
17:00:07 <dustymabe> #startmeeting fedora_atomic_wg
17:00:07 <zodbot> Meeting started Wed Jul 12 17:00:07 2017 UTC.  The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:07 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:07 <zodbot> The meeting name has been set to 'fedora_atomic_wg'
17:00:13 <dustymabe> #topic roll call
17:00:16 <yzhang> .hello yzhang
17:00:16 <dustymabe> .hello dustymabe
17:00:16 <zodbot> yzhang: yzhang 'Yu Qi Zhang' <jzehrarnyg@gmail.com>
17:00:19 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
17:00:25 <sayan> .hello sayanchowdhury
17:00:25 <jdoss> .hello jdoss
17:00:25 <zodbot> sayan: sayanchowdhury 'Sayan Chowdhury' <sayan.chowdhury2012@gmail.com>
17:00:28 <zodbot> jdoss: jdoss 'Joe Doss' <joe@solidadmin.com>
17:00:29 <tibbs> Howdy.
17:00:32 <ksinny> .hello sinnykumari
17:00:33 <zodbot> ksinny: sinnykumari 'Sinny Kumari' <ksinny@gmail.com>
17:00:36 <stefw> .hello stefw
17:00:37 <zodbot> stefw: stefw 'Stef Walter' <stefw@redhat.com>
17:00:40 <tibbs> Oops, wrong channel.
17:00:41 <maxamillion> .hello maxamillion
17:00:42 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
17:00:59 <dustymabe> #chair yzhang dustymabe sayan jdoss walters ksinny stefw maxamillion jbrooks
17:00:59 <zodbot> Current chairs: dustymabe jbrooks jdoss ksinny maxamillion sayan stefw walters yzhang
17:00:59 <strigazi> .hello Spyros Trigazis
17:01:00 <zodbot> strigazi: Sorry, but you don't exist
17:01:12 <strigazi> .hello strigazi
17:01:13 <zodbot> strigazi: strigazi 'Spyros Trigazis' <strigazi@gmail.com>
17:01:15 <dustymabe> #chair strigazi
17:01:15 <zodbot> Current chairs: dustymabe jbrooks jdoss ksinny maxamillion sayan stefw strigazi walters yzhang
17:01:19 <bowlofeggs> suuup
17:01:28 <dustymabe> bowlofeggs: o/
17:01:31 <jdoss> Ahoy
17:01:38 <jbrooks> .fas jasonbrooks
17:01:38 <zodbot> jbrooks: jasonbrooks 'Jason Brooks' <JBROOKS@REDHAT.COM>
17:01:40 <dustymabe> #chair bowlofeggs
17:01:40 <zodbot> Current chairs: bowlofeggs dustymabe jbrooks jdoss ksinny maxamillion sayan stefw strigazi walters yzhang
17:01:47 <dustymabe> #chair rubao
17:01:47 <zodbot> Current chairs: bowlofeggs dustymabe jbrooks jdoss ksinny maxamillion rubao sayan stefw strigazi walters yzhang
17:01:54 <dustymabe> #chair scollier
17:01:54 <zodbot> Current chairs: bowlofeggs dustymabe jbrooks jdoss ksinny maxamillion rubao sayan scollier stefw strigazi walters yzhang
17:02:00 <jberkus> .hello jberkus
17:02:00 <zodbot> jberkus: jberkus 'Josh Berkus' <josh@agliodbs.com>
17:02:12 <dustymabe> #chair jberkus
17:02:12 <zodbot> Current chairs: bowlofeggs dustymabe jberkus jbrooks jdoss ksinny maxamillion rubao sayan scollier stefw strigazi walters yzhang
17:02:12 <rubao> .hello rubao
17:02:17 <zodbot> rubao: rubao 'rubao' <rubao.net@hotmail.com>
17:02:18 <dustymabe> good turnout today
17:02:20 <scollier> .hello scollier
17:02:21 <zodbot> scollier: scollier 'Scott Collier' <emailscottcollier@gmail.com>
17:02:40 <scollier> dustymabe, that ping helped :)
17:03:01 <dustymabe> roshi around today?
17:03:21 <dustymabe> he was been quiet lately, but now that f26 is out i think he should perk back up a bit
17:03:29 <dustymabe> s/was/has
17:03:50 <dustymabe> kushal around?
17:04:02 <dustymabe> oh well - let's get started
17:04:08 <dustymabe> #topic previous meeting action items
17:04:22 <jdoss> I have to jet to another meeting in 15 but I wanted to ask if it was OK to open an issue to consider getting Fedora 26 added to Google Compute Engine's official images since CentOS and RHEL images are present already.
17:04:24 <miabbott> .hello miabbott
17:04:25 <zodbot> miabbott: miabbott 'Micah Abbott' <miabbott@redhat.com>
17:04:54 <sayan> jdoss: that's something I wanted to start the discussion in the open floor
17:04:55 <dustymabe> jdoss: I think it's something we want to do - there may be a ticket for it already, but if not feel free to open one
17:05:03 <jdoss> If the answer is yes I will open an issue with the steps I used to get it going for my own needs
17:05:14 <dustymabe> ok here are the previous meeting action items
17:05:17 <dustymabe> * maxamillion roshi to come up with guidelines for meeting quorum for
17:05:18 <dustymabe> the atomic working group
17:05:20 <dustymabe> * dustymabe to propose a VFAD for wiki and docs discussion
17:05:22 <dustymabe> * maxamillion to look at logs from last meeting and add summary of our
17:05:24 <dustymabe> discussion to ticket https://pagure.io/atomic-wg/issue/284
17:05:26 <dustymabe> * maxamillion to release the fedora-minimal base image with next
17:05:28 <dustymabe> container release: https://pagure.io/atomic-wg/issue/290
17:05:30 <dustymabe> * dustymabe to update the kubernetes thread with tracking options:
17:05:32 <dustymabe> https://pagure.io/atomic-wg/issue/287
17:05:34 <dustymabe> * dustymabe to open a ticket for cloud/atomic confusion
17:05:48 <dustymabe> #info dustymabe proposed a VFAD https://pagure.io/atomic-wg/issue/294
17:05:58 <maxamillion> so for https://pagure.io/atomic-wg/issue/290 ... I commented, but I didn't really sum up what was said in the meeting, which I completely forgot about
17:06:22 <dustymabe> #info dustymabe added update to kubernetes thread: https://pagure.io/atomic-wg/issue/287#comment-447645
17:06:27 <maxamillion> because I'm just bad at that lately
17:06:38 * maxamillion can't seem to keep his head on straight
17:07:00 <dustymabe> #info dusty opened a ticket regarding mailing list/irc channel for atomic/cloud working group: https://pagure.io/atomic-wg/issue/295
17:07:02 <jberkus> dustymabe: I'm triple-timing until :30, will be more responsive at that point
17:08:14 <dustymabe> maxamillion: you mean for #284?
17:08:35 <maxamillion> dustymabe: I do
17:08:41 <dustymabe> ok
17:08:49 <dustymabe> for #290 we are done - its in the registry
17:08:51 <dustymabe> right?
17:09:03 <maxamillion> dustymabe: yes
17:09:14 <dustymabe> #info fedora-minimal container image is in the fedora registry now https://pagure.io/atomic-wg/issue/290
17:09:34 <dustymabe> ok the only item left is:  * maxamillion roshi to come up with guidelines for meeting quorum for the atomic working group
17:09:43 <dustymabe> re-action I assume?
17:10:59 <maxamillion> dustymabe: yeah, I'll try and track him down soon
17:11:18 <dustymabe> #action maxamillion roshi to come up with guidelines for meeting quorum for the atomic working group
17:11:25 <dustymabe> ok moving on to meeting items
17:11:44 <dustymabe> #topic Start using new mailing list/IRC channel around f26 release time
17:11:52 <dustymabe> #link https://pagure.io/atomic-wg/issue/295
17:12:15 * dustymabe will give people some time to read the ticket
17:12:23 <dustymabe> TL;DR - let's start using
17:12:31 <sayan> the atomic mailing list?
17:12:31 <dustymabe> 'atomic' named things for this working group
17:12:58 <dustymabe> atomic mailing list - fedora-atomic irc channel - atomic calendar entries in fedocal
17:13:03 <dustymabe> atomic namespace on the wiki
17:13:20 <dustymabe> etc..
17:13:22 <yzhang> makes sense
17:13:29 <jbrooks> I plus one-d
17:13:44 <dustymabe> yes - please add your thoughts to the ticket
17:13:46 <ksinny> +1
17:14:13 <maxamillion> +1
17:14:28 <dustymabe> cloud* stuff won't go away - just will now be used more eclusively for cloud base image
17:14:34 <dustymabe> and related issues
17:14:35 <sayan> what happens to the cloud channel?
17:14:44 <dustymabe> sayan: ^^
17:14:45 <sayan> ah ok
17:14:47 <sayan> Yeah
17:14:50 <gholms> Is there too much traffic in the existing places?
17:14:59 <gholms> I mean, why bother?
17:15:02 <dustymabe> gholms: it's mostly just confusing to new users
17:15:06 <walters> this proposes adding #fedora-atomic in addition to #atomic ?
17:15:26 <dustymabe> walters: yeah - i teetered on that one
17:15:36 <dustymabe> could be swayed
17:15:43 <stefw> indeed, why not just #atomic ?
17:15:53 <gholms> dustymabe: A redirect can solve that one.
17:16:00 <dustymabe> i feel like there are a lot of fedora specific discussions that happen in #fedora-cloud
17:16:10 <dustymabe> that #atomic channel won't care much about
17:16:24 <gholms> Is it okay for people to go to the Atomic upstream channel with Fedora traffic?
17:16:39 <gholms> That'd be the main argument for or against that.
17:16:43 <dustymabe> right
17:16:45 <stefw> isn't the Fedora version of atomic *the* upstream Atomic Host project?
17:16:53 <jzb> stefw: +1
17:16:58 <jzb> or should be anyway :-)
17:17:00 <maxamillion> stefw: depends on who you ask
17:17:04 <dustymabe> well - i think there is grey area
17:17:11 <dustymabe> #atomic is more a collection of upstream projects
17:17:18 <dustymabe> including atomic CLI
17:17:19 <gholms> If it is then the messaging needs an awful lot of work.
17:17:19 <stefw> well for the host as a whole
17:17:21 <miabbott> "project atomic" adds to the confusion
17:17:26 <dustymabe> rpm-ostree ostree blah blah
17:17:29 <stefw> it's already confusing it enough as it is ... i'd say if you're changing things around ... give a nod toward simplicity and consistency
17:17:52 <jdoss> It was always confusing to me on all the atomic chatter in the #fedora-cloud channel since I haven't drank the atomic koolaid yet
17:17:55 <dustymabe> stefw: suggestion?
17:17:56 <jbrooks> I wouldn't mind using #atomic for more fedora atomic (and centos atomic) stuff
17:18:12 <dustymabe> jdoss represents an outside user
17:18:14 <jberkus> If we're changing channels, it would be better to just use #atomic
17:18:15 <stefw> #atomic ... or if you don't want host specific stuff there then #atomic-host
17:18:21 <stefw> jberkus ++
17:18:24 <dustymabe> so he is a good example of someone that would be confused by some of our choices
17:18:34 <walters> i think i'm +1
17:18:39 <jberkus> the fact that we'll be mixing with cockpit and Atomic CLI etc. is a feature, not a bug
17:18:45 <dustymabe> someone want to lay out a proposal?
17:18:46 <maxamillion> miabbott: +1
17:18:50 <jdoss> I get it now but any new user that is using Fedora Cloud images is going to be hella confused if they join
17:18:51 <gholms> I just want to make sure people will be okay with having every distro using the same channel to coordinate their atomic stuff.
17:19:02 <maxamillion> jberkus: +1
17:19:07 <dustymabe> jdoss: #fedora-cloud will still be there
17:19:10 <dustymabe> for the cloud base image
17:19:28 <jdoss> sweet!
17:19:30 <dustymabe> here is a question
17:19:38 <dustymabe> walters: - if the endless os guys want to talk about ostree stuff
17:19:42 <dustymabe> where would that happen?
17:19:44 <miabbott> #ostree
17:19:46 <walters> we chat today in #ostree
17:19:51 <dustymabe> ok
17:20:00 * dustymabe adds that to his list of channels
17:20:19 <dustymabe> yeah - it's all a balance
17:20:30 <jberkus> I'd also like to discuss the tradeoffs of using atomic-devel@projectatomic.io instead
17:20:53 <dustymabe> jberkus: i think our list is way too chatty for that
17:21:10 <jzb> dustymabe: which list?
17:21:18 <dustymabe> cloud@lists.fp.o
17:21:26 <dustymabe> which would become atomic@lists.fp.o
17:21:32 <jberkus> well, the only part that's chatty is the automated notices
17:21:40 <jzb> dustymabe: and is there really a lot of conversation on that list that isn't germane to atomic-devel?
17:21:42 <jberkus> otherwise, the cloud@ list is rarely used
17:21:42 <dustymabe> but that's where the discussion is happening
17:21:58 <jberkus> jzb: there really isn't
17:22:18 <jberkus> Fedora Atomic discusses things in two locations: this channel, and Pagure issues
17:22:18 * jzb will get the same amount of traffic either way
17:22:26 <dustymabe> i'm just saying that the mails from our discussions in our issue tracker would dominate that mailing list
17:22:27 <jbrooks> I like more atomic alignment
17:22:37 <jbrooks> Does the issue tracker go to the list?
17:22:42 <dustymabe> jzb: i'm not worried about people who are subscribed to both lists
17:22:50 <gholms> If the projectatomic.io site wasn't so unquestionably not a Fedora-focused page I could see that.
17:22:51 <jberkus> jbrooks: I think so?
17:22:57 * stefw notes for reference that we talk about lots of distro specific and packaging Cockpit stuff in #cockpit. And Cockpit is also branded, customized, and heavily tested, and delivered per distro
17:22:58 <dustymabe> jberkus: yes
17:22:59 <sayan> jbrooks: yes
17:23:10 * gholms ponders
17:23:17 <jzb> what about bringing Project Atomic under Fedora?
17:23:27 <jzb> officially
17:23:39 <gholms> That'd fix that.
17:24:05 <jberkus> centos atomic is still a thing.  So is Red Hat Atomic
17:24:25 <stefw> jberkus, but making one the upstream has real merit
17:24:31 <dustymabe> yeah honestly I don't really care about all that stuff (who is under what umbrella etc)
17:24:32 <jzb> jberkus: both downstreams (more or less) of Fedora Atomic
17:24:35 <gholms> Yeah, but those aren't relevant to Fedora.
17:24:54 <jberkus> however, realistically, there aren't many members of any of the associated atomic projects who aren't also on this team
17:25:02 * dustymabe notes it's interesting this topic sparked this long of a discussion
17:25:08 <jberkus> and it's *way* easier for new users to have Just One List
17:25:14 <dustymabe> jberkus: i would say that's not the case
17:25:15 <jbrooks> I don't see the value in folding project atomic into fedora, but I see the value in using the project atomic ml for us
17:25:18 <jzb> dustymabe: of course, naming discussions always do ;-) Also green.
17:25:18 <jdoss> any new outside users that might help out with stuff might be confused on what project is what
17:25:23 <jdoss> since we are not part of this team
17:25:31 <dustymabe> so we often don't have container team members attend our meeting
17:25:46 <dustymabe> also this is the first time we've had someone from cockpit AFAIK
17:26:11 <dustymabe> which is ok - just saying it has traditionally been more focused on "host"
17:26:14 <maxamillion> just as a side note, if Project Atomic claimed to be moved under the Fedora umbrella ... would there need to be a formal proposal to Fedora Council as well as some group of Project Atomic folks? (I don't know the project's governing structure)
17:26:26 <jbrooks> etc etc etc yeah
17:26:35 <jbrooks> don't see the value in all that
17:26:41 <jberkus> maxamillion: yes, there would
17:26:46 <maxamillion> rgr
17:26:50 <dustymabe> ok let me try to summarize
17:27:03 <jbrooks> the prd stuff was painful enough ;)
17:27:22 <dustymabe> - 1 - some concern over creating new lists/places - why not use atomic-devel mailing list and #atomic irc channel
17:27:51 <dustymabe> - 2 - mostly because there are things that are very fedora specific happening in #fedora-cloud and on the cloud mailing list and we don't want to flood the channel
17:28:30 <dustymabe> - 3 - atomic-devel is meant for all project atomic projects and we could cause too much chatter there
17:28:43 <dustymabe> so that's the gist, from what I gather?
17:28:47 <jbrooks> There's no need to fwd all issue tracker items to the list, though
17:28:51 <jberkus> 4 - chatter is really just automated notices, so we could handle those separately
17:28:53 <jbrooks> And who's saying #3?
17:29:13 <dustymabe> jbrooks: I guess me for #3
17:29:35 <jbrooks> https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-June/thread.html
17:29:37 <dustymabe> I would like to keep the pagure emails going to a list
17:29:42 <jbrooks> not a ton on there
17:30:15 <dustymabe> jbrooks: yeah - we would probably double or triple the traffic
17:30:19 <gholms> I've got one:  it would be the only Fedora product that doesn't use Fedora lists/channels/etc.
17:30:37 <maxamillion> gholms: +1
17:30:50 <jbrooks> Yeah, fine, we can just do new list / irc
17:30:55 <dustymabe> anyone want to talk about notifications going to a list?
17:30:57 <jbrooks> I'm going to get them both anyway
17:31:06 <jbrooks> I think it's excessive
17:31:13 <jbrooks> I've been thinking about filtering some of it
17:31:14 <maxamillion> which is or isn't a real concern, I don't really know ... Atomic has broken a lot of the mold already, but it's a fair point to make note of
17:31:16 <dustymabe> i like them going to a list because you get a global view of what  everyone is seeing
17:31:18 <gholms> dustymabe: I was going to ask that we kick them out of the cloud list anyway.
17:31:34 <dustymabe> it doesn't depend on each persons individual settings
17:31:38 <gholms> The notifications
17:31:41 <dustymabe> if they are on the list, they are seeing the emails
17:31:50 <dustymabe> gholms: yes, when we move they would move with it
17:31:54 <jbrooks> I figure, if I want to be subbed to an issue, I'll sub myself
17:32:02 <dustymabe> gholms: note that the cloud-sig repo will still send notificatiosn to the list
17:32:33 <gholms> dustymabe: Well, I was going to ask that it not anyway.  ;)
17:32:47 <dustymabe> another reason to send notifs to the list is because that is how we communicate some changes (i know we should send formal emails more often, but it doesn't always happen)
17:32:47 <jbrooks> So I'm +1 to a new fedora-atomic irc and ML
17:33:10 <gholms> There is so much pagure mail on the list that I just ignore it.
17:33:25 <jbrooks> yep
17:33:32 <dustymabe> gholms: right - but you are more interested in cloud base image aren't you?
17:33:44 <maxamillion> same, I filter all pagure email out and just ignore it ... I'm subscribed to like 20 repos, the noise is insane
17:34:00 <dustymabe> maxamillion: i just filter the lists to separate folders
17:34:05 <jberkus> I'm -1 on a new IRC channel, but willing to discuss list particulars
17:34:07 <dustymabe> and then run through them
17:34:35 <jbrooks> jberkus, Ah, yeah, that'd be something, a split
17:34:41 <gholms> dustymabe: I'm not sure what that has to do with the mailing list.
17:34:52 <jbrooks> Really, I'd love to make #atomic my main hangout for fah and cah
17:35:01 <dustymabe> gholms: in other words our atomic tracker has been sending emails to the cloud list
17:35:04 <dustymabe> so sure, you don't care about them
17:35:11 <jberkus> materially, many devs use the two channels interchangeable
17:35:13 <dustymabe> so that is probably why you ignore them
17:35:19 <jberkus> so they might as well be one channel
17:36:06 <dustymabe> ok i think we've talked about this enough and no real concensus at this point
17:36:13 <gholms> Heh
17:36:16 <dustymabe> please add concerns/thoughts to the ticket
17:36:18 <jberkus> continue discussion on ticket
17:36:22 <dustymabe> and vote if you will
17:36:31 <dustymabe> add proposals etc..
17:37:19 <dustymabe> #info lots of good discussion here - no real concensus (spelling?) - continue discussion in ticket https://pagure.io/atomic-wg/issue/295
17:37:28 <dustymabe> #topic Atomic Host images omit many common locales that all other flavors include
17:37:34 <dustymabe> #link https://pagure.io/atomic-wg/issue/282
17:37:51 <dustymabe> so jlebon worked on some items around this ticket
17:38:10 <dustymabe> which should address adam's most pressing concerns
17:38:26 <dustymabe> He'll continue to update the ticket as things go
17:38:41 <dustymabe> #info jlebon worked on some items around this ticket which should address adam's most pressing concerns
17:39:07 <dustymabe> #topic Delivering Atomic Host via CI/CD pipeline
17:39:12 <dustymabe> #link https://lists.fedoraproject.org/archives/list/cloud@lists.fedoraproject.org/message/ZS7XY7NHXGGOPOB2YKNQUWSDUGCMYIL5/
17:39:18 <dustymabe> #link https://fedoraproject.org/wiki/Objectives/Continuous_Integration_and_Delivery_of_Fedora_Atomic_Host
17:39:22 <dustymabe> stefw: take it away
17:39:24 <stefw> so the basic elevator pitch:
17:39:47 <stefw> We compose atomic host artifacts exactly as run by users
17:39:52 <stefw> test those artifacts during integration testing
17:40:07 <stefw> provide feedback to packagers, and optionally gate packagers, gate inclusion in the host based on that testing
17:40:19 <stefw> and then deliver exactly those bits (qcow2, ostree, ami) as tested
17:40:45 <stefw> This would be an Objective for Fedora
17:41:14 <stefw> it will benefit the quality of Fedora overall, by gating broken package changes, and providing feedback from testing directly to those making changes to packages.
17:41:21 <stefw> but it's focus and scope is specifically on Atomic Host
17:41:32 <maxamillion> I'm in
17:41:34 <stefw> Atomic Host is based on technology (OSTree) that enables this
17:41:36 <dustymabe> stefw: one question - from a packager perspective - this would integrate with bodhi?
17:41:41 <stefw> but we never went the full way to actually pull this off
17:41:41 <jbrooks> sounds awesome
17:41:49 <stefw> dustymabe, I believe it would
17:41:53 <stefw> i believe there would be two levels of gating
17:42:04 <stefw> one at the package level ... where a packager receives feedback to say the package is broken for Atomic Host
17:42:07 <walters> so in this, do i still have to beg for karma in bodhi too?
17:42:11 <walters> e.g. https://bodhi.fedoraproject.org/updates/FEDORA-2017-ad3832d881
17:42:14 <gholms> Nice
17:42:14 <stefw> walters, i don't think so
17:42:29 <walters> as far as process goes bodhi is the most broken IMO
17:42:31 <stefw> a packager receives feedback ... that a broken change (ie: NVR) is not going to get into Atomic Host
17:42:43 <stefw> pingou has done work to make bodhi approval automatic based on integration testing
17:42:47 <stefw> that is not functional yet
17:42:50 <walters> but it would go into buildroots/workstation?
17:43:03 <stefw> walters, i believe the solution there is related to modularity
17:43:11 <stefw> but we may choose another path
17:43:13 <stefw> basically the concept that
17:43:24 <stefw> a specific set of packages (NVRs) known to be good go into an Atomic Host compose
17:43:33 <stefw> some of those packages are shared by other buildroots, and general delivery of Fedora
17:43:36 <stefw> or other modules
17:43:37 <jberkus> stefw: we'll need to improve our testing
17:43:43 <gholms> Interesting idea
17:43:46 <stefw> jberkus, certainly
17:43:52 <stefw> so porting tests to support this effort is underway
17:43:59 <jberkus> is roshi around?
17:44:10 <stefw> Early this year Fedora approved the concept of placing tests in dist-git
17:44:11 <dustymabe> jberkus: I pinged him earlier, he may be AFK
17:44:18 <stefw> to be changed along with packages and/or modules (for higher level tests)
17:44:27 <jberkus> stefw: one thing we'd really need is end-to-end testing for either FAO or Fedora+Kube, ideally both
17:44:30 <stefw> and there is active movement to create or port such tests
17:44:41 <stefw> jberkus, the current PAPR tests are a great place to start for that
17:44:55 <stefw> staging Pagure instance for tests ready to go into dist-git
17:44:56 <stefw> https://upstreamfirst.fedorainfracloud.org/
17:45:09 <dustymabe> jberkus: yeah - i think system level tests for atomic host + openshift and/or kube are important
17:45:11 <stefw> pingou and threebean puiterwijk are working on getting Pagure in front of dist-git
17:45:25 <stefw> dustymabe, indeed, and they fit really well in the concept of a module in dist-git that defines Atomic Host
17:45:25 <dustymabe> +1 for pagure on dist-git - can't wait
17:45:37 <stefw> that module which defines Atomic Host, would contain broader Atomic Host specific tests
17:45:42 <dustymabe> stefw: a few more questions
17:45:44 <stefw> so the story is rather consistent end to end
17:45:49 <ksinny> stefw: We are trying to make Fedora Atomic Host available for aarch64 and ppc64le as well along with x86_64 from F27 onwards. I am hoping that these arches will get benifited too with Fedora Atomic CI
17:46:04 <maxamillion> ksinny: +1
17:46:05 <dustymabe> is anyone working on this now (i.e. the fedora build process/tooling) and how atomic host is composed
17:46:14 <dustymabe> and how do we start working with them
17:46:23 <stefw> dustymabe, if you look at the Objective there are links to some initial work done
17:46:26 * stefw gets a link
17:46:37 <stefw> whoops it's on this landing page
17:46:43 <stefw> https://fedoraproject.org/wiki/CI
17:47:14 <stefw> in addition on that landing page there are links to more information about test layout and test invocation
17:47:20 <stefw> aswell as links to the pipeline
17:47:39 <stefw> lastly Fedora has put together an outline for how a CI/CD system and pipeline would interact with Fedora: https://fedoraproject.org/wiki/Fedora_requirements_for_CI_and_CD
17:47:45 <dustymabe> #info some people have been looking at fedora build process/tooling for fedora atomic and have some initial work links from https://fedoraproject.org/wiki/CI
17:48:02 <stefw> ksinny, the multi arch question is a good one
17:48:03 <stefw> it's two fold
17:48:08 <stefw> first of all when you run CI at scale
17:48:24 <stefw> it's pretty clear that you get 95% of the way (or more) by running continuous integration testing on one or two architectures
17:48:34 <stefw> usually those are very different architectures such as 64-bit x86_64 and 32-bit i386
17:48:47 <stefw> it is often the case that you make up for the other architectures during the delivery phase
17:49:03 <ksinny> stefw: I assume that CI work  will use CentOS infra. If yes, we are working on adding ppc64le hardware at least for the start
17:49:07 <stefw> but yes, these are aspects that that can change over time
17:49:28 <stefw> ksinny, yes, that's good news and will be a big help
17:49:43 <ksinny> stefw: good to know :)
17:49:49 <stefw> but my point is that it's possible not to have literally everything duplicated, and still have an effective CI system
17:50:01 <stefw> a CI system will run thousands of composes, many will fail in testing, and be thrown away
17:50:14 <dustymabe> stefw: I know you mentioned pingou, is there anyone else from the fedora community we can collaborate with
17:50:15 <stefw> it's possible to do that in an intelligent way, by maxing out x86_64 hardware, before composing elsewhere
17:50:29 <stefw> there's a mailing list here: ci@lists.fedoraproject.org
17:50:36 <stefw> in addition, there are people like:
17:50:51 <stefw> merlin mathesius shepherding the porting of tests for core packages
17:50:55 <dustymabe> #info there is a mailing list for the ci efforts in Fedora, that will initially focus on the set of packages in atomic host: ci@lists.fedoraproject.org
17:51:15 <stefw> tim flink running the staging area for tests being ported from Red Hat
17:51:35 <dustymabe> does the group associated with that list have a pagure issue tracker or irc channel?
17:51:38 <sayan> stefw: is there a IRC channel where you do the discussions?
17:51:45 <stefw> there is an IRC channel #fedora-ci
17:51:56 <stefw> there is no pagure issue tracker yet
17:52:14 <stefw> i'm wondering if using the Fedora Atomic/Cloud issue tracker would be a good thing for now?
17:52:26 <stefw> given that this effort is very much focused on Atomic Host for the time being
17:52:33 <dustymabe> stefw: i thik it depends on how chatty it is
17:52:45 <stefw> indeed.
17:52:54 <dustymabe> would be cool if we could configure notifications individually based on labels
17:53:03 <stefw> i imagine it will start off steadily and as CI in Fedora goes beyond Atomic Host, we could probably outgrow the Atomic Host specific tracker
17:53:19 <dustymabe> we can start there and then grow
17:53:50 <dustymabe> starting there might be a good way to raise awareness too
17:54:04 <dustymabe> cool ok, anything you need from the working group for now?
17:54:06 <stefw> So i'd like to raise this an a Fedora Objective
17:54:08 <dustymabe> action items, etc?
17:54:28 <stefw> And I can't just do that on my own
17:54:30 <dustymabe> also, anyone have any concerns about an effort like this?
17:54:37 <stefw> so I'd like some general feeling on the concerns around this effort
17:54:44 <stefw> and who would back it, who needs more time to think
17:54:45 <stefw> etc.
17:54:45 <dustymabe> stefw: jinx
17:54:50 <stefw> :D
17:55:21 <maxamillion> I'm in, I love the idea ... I'm concerned about some of the implementation details, it seems like a lot of work and it touches a lot of moving parts but ultimately I'm in
17:55:31 <dustymabe> stefw: honestly I think the idea as a whole sounds good - but we don't have much to go on as far as specifics go
17:55:47 <dustymabe> i think most of the people here agree more testing would be lovely :)
17:55:50 <stefw> maxamillion, indeed, you're absolutely right, touches a lot of things, requires some change of mindset too, tough stuff
17:55:59 <walters> if i can stop manually editing specfiles and manually initializing kerberos to tell koji i did a git push and then manually telling bodhi i built an rpm...i am really excited
17:55:59 <miabbott> there is a lot to digest in the wiki pages that stefw linked
17:56:05 <maxamillion> stefw: indeed
17:56:06 <walters> if I have to do that *and* watch something else, that kind of sucks
17:56:06 <dustymabe> and this will take a lot of collaboration with releng/infra
17:56:17 <stefw> dustymabe, indeed, i left out many of the imagined or prototyped details
17:56:25 <maxamillion> miabbott: +1 - I had already read it, I saw mention of it on th emailing list
17:56:39 <stefw> because I didn't want it to seem like this was all already decided ... therhe are lots of ideas about how things would work
17:56:56 <stefw> should I add an ideas section to the Objective? things that are not set in stone, but help paint a fuller picture?
17:57:20 <dustymabe> stefw: sure, sounds great
17:57:35 <dustymabe> stefw: thanks for bringing this to the group
17:57:36 <maxamillion> dustymabe: note that members of both fedora and centos infra are involved in the CI Objective
17:57:51 <dustymabe> maxamillion: cool
17:58:16 <dustymabe> ok anyone else with anything
17:58:26 <stefw> wait so, one qu estion
17:58:41 <stefw> as far as raising this as Fedora Objective, are we:
17:58:44 * pingou can neither confirm nor deny knowin dustymabe
17:58:49 <stefw> 1. Good to go now
17:58:52 <maxamillion> pingou: :D
17:58:54 <stefw> 2. Wait for further discussion on the mailing list?
17:59:27 <jberkus> I'm +1, but very little of the work will be mine
17:59:34 <dustymabe> stefw: i think i'm all for making a draft proposal and then making edits
17:59:38 <stefw> In other words, i'm ready to take the next step, but would love to work you folks further on this, as necessary.
17:59:39 <dustymabe> also timelines will be interesting
17:59:52 <stefw> indeed
18:00:05 <stefw> so part of what may come out of this work is the idea of a single stream of atomic host
18:00:07 <dustymabe> so dates for when this will happen need to be figured out, but the amount of work to be done hasn't quite been quantified
18:00:11 * ksinny is interested as well
18:00:23 <maxamillion> dustymabe: +1
18:00:24 <stefw> perhaps less tied to Fedora releases ... but that's a bit hazy in the future ...
18:00:50 <miabbott> dustymabe: i would quantify the amount of work as 'significant'  :)
18:00:53 <dustymabe> stefw: yeah, breaking out of the fedora 'releases' could be something we do in the future
18:01:07 <dustymabe> but there is a lot of work we get for free by being tied to them right now
18:01:15 <dustymabe> so pros/cons would have to be weighed
18:01:26 <stefw> dustymabe, indeed, certainly not the first thing to tackle
18:01:27 <dustymabe> it's a love/hate relationship :)
18:01:39 <dustymabe> ok we're over time
18:01:49 <dustymabe> but I want to see if anyone had anything pressing for open floor
18:01:52 <dustymabe> #topic open floor
18:02:04 <dustymabe> - FYI Fedora 26 Atomic host first release was yesterday
18:02:17 <sayan> I'll probably ask about GCE in the cloud channel
18:02:39 <dustymabe> - FYI - please read the announcement from projectatomic blog: http://www.projectatomic.io/blog/2017/07/fedora-atomic-26-release/
18:02:48 <dustymabe> #info Fedora 26 Atomic host first release was yesterday please read the announcement from projectatomic blog: http://www.projectatomic.io/blog/2017/07/fedora-atomic-26-release/
18:03:04 <dustymabe> anyone else with anything?
18:03:16 <gholms> Thanks, everyone  :)
18:03:23 <dustymabe> 3...
18:03:29 <dustymabe> 2..
18:03:29 <stefw> ô/
18:03:33 <dustymabe> 1.
18:03:36 <dustymabe> #endmeeting