20:00:08 <stickster> #startmeeting Atomic/Infrastructure
20:00:08 <zodbot> Meeting started Tue Aug 19 20:00:08 2014 UTC.  The chair is stickster. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:08 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:00:28 <stickster> .hellomynameis pfrields
20:00:29 <zodbot> stickster: pfrields 'Paul W. Frields' <stickster@gmail.com>
20:00:35 <walters> .hellomynameis walters
20:00:36 <zodbot> walters: walters 'Colin Walters' <walters@redhat.com>
20:01:03 <nirik> .hellomynameis kevin
20:01:05 <zodbot> nirik: kevin 'Kevin Fenzi' <kevin@scrye.com>
20:01:11 <stickster> #chair walters
20:01:11 <zodbot> Current chairs: stickster walters
20:01:14 <stickster> #chair nirik
20:01:14 <zodbot> Current chairs: nirik stickster walters
20:02:03 <stickster> dgilmore: oddshocks: number80: ping
20:02:19 * jbrooks lurks
20:02:21 <walters> so my high level understanding right now is that the "alt" composes of rawhide are mostly blocked on atomic development of new image generation code and deployment, which we're making progress on
20:02:23 <stickster> #info jzb is traveling to LinuxCon so stickster is subbing for him
20:02:29 <stickster> #topic Status check
20:02:29 * misc lurk too
20:02:33 <walters> then hopefully some of that can be transferable to mainline rel-eng
20:02:53 * dustymabe lurking
20:03:51 <walters> the other topics i had were:
20:03:56 <walters> * mirrormanager/metalinks
20:04:01 <walters> * official rel-eng composes
20:04:40 <stickster> #info Alt composes of rawhide blocked on Atomic generating and deploying new images --> in progress
20:05:22 <walters> do we have dgilmore?
20:05:38 <stickster> He hasn't responded to my ping.
20:06:14 <nirik> he did say he was working on it eariler today...
20:06:17 <nirik> but I don't know status
20:06:44 <stickster> walters: Can I plunge into MM topic?
20:06:48 <walters> yep
20:06:51 <stickster> #topic MirrorManager/metalinks
20:06:59 <walters> ok, on the mirrormanager/ostree side, I've made significant progress on a metalink implementation, and should be ready to test it out when we have content synced to mirrors and a metalink.xml generated
20:07:35 <stickster> oddshocks and pingou are working on a FAD for MirrorManager as well, but I think that has more to do with a port (MM2), and does not block this at all
20:07:37 <nirik> awesome
20:07:51 <nirik> yeah, although we need to make sure any work here also works in MM2.
20:07:58 <stickster> nirik: Agreed
20:08:36 <stickster> #info walters made progress on metalink implementation, ready to test once content is sync'd to mirrors and metalink.xml is generated -- need to make sure this works in MM2 later as well.
20:08:56 <pingou> I don't think (atm) MM2 will work much differently than the way MM1 works
20:09:07 <pingou> (if that makes sense)
20:10:14 <nirik> walters: once we have a test compose we can probibly test it there.
20:10:20 <walters> one detail is at the moment the file is named "summary", but if necessary we can just symlink repomd.xml -> summary and be done with it
20:11:26 <stickster> walters: So do I understand correctly you're waiting now for someone to finish the task of generating the metalink.xml ?
20:11:51 <walters> stickster, yes, which is in turn waiting on composes and mirroring them I believe
20:12:07 <stickster> Oh, I just realized why oddshocks isn't here -- he's at New Hire Orientation at Red Hat @RDU.
20:12:24 <stickster> Ah, the wheel goes 'round... ;-)
20:13:16 <stickster> #info oddshocks in new hire orientation 18-19 Aug, then relocating to SEA but should be online afternoon/evening-ish while traveling
20:13:18 <agrimm> oh, wish I had known to go look for him yesterday
20:13:38 <agrimm> they are out of NHO by now
20:13:44 <agrimm> he's probably at the bar.  :)
20:13:48 <stickster> Yeah, and in his case probably on his way to the airport
20:14:01 <stickster> I'll make sure he sees these notes and checks in on-list
20:14:28 <stickster> #action stickster remind oddshocks to read log and check in on-list
20:15:33 <stickster> walters: anywhere else you're blocked or need input/
20:16:59 <walters> i think the main block for f21 is the official compose
20:17:55 <walters> geppetto, agrimm: what do you guys think?
20:18:31 <stickster> #topic rel-eng compose
20:18:44 <walters> there is a whole thread where we're going to have to figure out how to make what's currently on alt more official as we have more deliverables like an installer ISO as well as a live payload
20:19:34 <stickster> walters: what list is that on?
20:19:43 * dgilmore was afk
20:19:44 * jzb slips in quietly
20:19:48 <walters> but let's just figure out how to get the f21 cloud image out
20:19:50 <agrimm> walters, you are probably correct
20:19:51 <walters> (first)
20:19:54 <stickster> ha, nice try jzb, you said that out loud ;-)
20:20:02 <stickster> #chair dgilmore jzb agrimm
20:20:02 <zodbot> Current chairs: agrimm dgilmore jzb nirik stickster walters
20:20:04 <dgilmore> ive sorted out making cloud images again
20:20:08 <jzb> d'oh
20:20:15 <dgilmore> i need to do some test to install a atomic tree now
20:20:19 <agrimm> dgilmore, aweseome
20:20:32 <stickster> #info dgilmore has sorted out producing cloud images  \o/
20:20:54 <walters> stickster, i posted to atomic-devel an installer ISO a while back, we haven't yet published the live payload though
20:21:07 <geppetto> walters: I guess … obviously lots of other things, but I can't think of any that would block release.
20:21:11 <stickster> #action dgilmore test installation of an atomic tree
20:21:32 <dgilmore> I am hoping to get an actual TC done today
20:21:37 <jzb> walters: just FYI, if I didn't say so - I have tested that and it worked...
20:21:54 <jzb> walters: did a few installs to virt manager and also on an old ThinkPad.
20:22:25 <stickster> #info jzb tested the installer ISO posted to atomic-devel earlier, fwiw
20:22:33 <walters> jzb, cool, yeah have several confirmations, and i think we're in place to have them regularly updated for rawhide at least
20:23:04 <stickster> What needs to happen next?
20:23:12 <agrimm> walters, FWIW, I built a vagrant image today, but the network seems broken for some reason.  that's likely just a minor kickstart issue, though
20:23:21 <walters> ah yeah, and the vagrant box
20:24:59 <stickster> Let's record who's got the ball for the minutes ;-)
20:25:13 <stickster> What's the next step, and who's accountable
20:26:05 <stickster> walters: jzb: agrimm: dgilmore: ^
20:26:14 <walters> i think GPG is the other outstanding piece as far as rel-eng
20:26:14 <dgilmore> stickster: its all new to me
20:26:40 <dgilmore> walters: im waiting on your answer there if a second signed copy of the file is fine
20:27:20 <walters> the file = summary file (the target of the metalink?)
20:27:44 <dgilmore> walters: I have no idea as you really have not made it totally clear what needs signed
20:28:14 <stickster> Let's do that now then :-)
20:28:25 <dgilmore> walters: but for fedup we sign the .treeinfo file and provide teh signed copy as .treeinfo.signed
20:28:37 <walters> the ostree default model has commit objects which are immutable, and detached metadata which is mutable, and an array of gpg signatures is a potential component of the detached metadata
20:28:47 <dgilmore> I asked if we provide a .signed copy of the files that should be signed a few times now and have not gotten an answer
20:29:34 <stickster> OK, let's give the discussion a chance
20:30:06 <walters> we could probably support a separate inline signed copy of the summary file
20:30:24 <dgilmore> the original would be unmodified, but instread of an .asc which we do not do we would have a coplete second copy that has an inline signature
20:30:28 <dgilmore> complete
20:30:52 <dgilmore> okay so lets do that
20:31:23 <walters> ok
20:31:57 <walters> (ok, with major reservations about having entirely different gpg signing codepaths between fedora/rhel)
20:32:18 <walters> but if that's what it takes to move forward, i can just do it
20:32:58 <stickster> #agreed atomic code will support a separate inline .signed copy of the summary file
20:33:07 <jzb> walters: is there anything that would prevent circling back and doing it in a way you prefer later, maybe for F22?
20:33:22 <stickster> #action dgilmore produce .signed file as needed
20:33:41 <stickster> #action walters support .signed file as agreed
20:34:21 <walters> jzb, yeah, hopefully we can revisit
20:34:32 <stickster> jzb: Yeah, that's what I was wondering too. I wouldn't think it's carved in stone, but obviously it would take someone working with the compose code to do it differently.
20:35:26 <jzb> cool
20:35:28 <stickster> Is there any other status we need to check today?
20:35:50 <walters> i think that's all for critical path of simply getting bits from A to B
20:36:00 <jzb> since I got here late, anything that I need to help with?
20:36:13 <jzb> or any tasks I was assigned as is the custom for people who're absent? :-)
20:36:25 <stickster> jzb: It seemed like all hard-tech stuff today, so yeah, we just put everything on your list
20:36:33 <jzb> fair enough
20:36:39 <stickster> :-)
20:36:50 <stickster> #topic AOB - All other business
20:37:10 <stickster> Is there anything else we need to cover?  Concerns, blockages, etc.?
20:37:14 <jzb> just fyi, I'm in Chicago for Linuxcon & doing a talk on Atomic later this week
20:37:23 <stickster> Wooo! Community building \o/
20:37:34 <walters> let's talk about rawhide in the next meeting?  (see above wrt installer ISO etc)
20:37:45 <jzb> so I'll be skimming the minutes and hope to give a status / roadmap update.
20:37:56 <misc> ( and number80 is doing a talk about atomic next monday )
20:38:19 <jzb> misc: CentOS Dojo?
20:38:22 <misc> jzb: yep
20:38:29 <misc> http://wiki.centos.org/Events/Dojo/Paris2014
20:38:37 <dustymabe> walters: jzb: anybody going to be in NYC anytime soon? would be nice to have a talk up here :)
20:38:52 <stickster> #action jzb Next meeting - discuss Rawhide status
20:38:56 <dgilmore> walters: we should talk about how you made that iso. as its not anything on the radar yet.
20:38:57 <number80> yup
20:39:13 <jzb> dustymabe: I don't have any plans atm but anything is possible.
20:39:31 <walters> dgilmore, https://git.fedorahosted.org/cgit/fedora-atomic.git/tree/compose/treecompose
20:40:02 <number80> I hope my atomic talk will get accepted for openworld forum too
20:40:28 <dustymabe> jzb: I'll cross my fingers
20:41:00 <stickster> Anything else for minutes before we close the meeting?
20:41:07 <stickster> we can chat in #atomic obviously ;-)
20:41:38 <stickster> #info rel-eng could use note on how walters made the installation ISO
20:41:41 <jbrooks> dgilmore: I think this is how the iso is made: https://lists.projectatomic.io/projectatomic-archives/atomic/2014-July/msg00002.html
20:41:58 <dgilmore> walters: okayt thats not at all how we would go about it.
20:42:10 <stickster> #undo
20:42:10 <zodbot> Removing item from minutes: INFO by stickster at 20:41:38 : rel-eng could use note on how walters made the installation ISO
20:42:15 * stickster RIF
20:42:16 <walters> dgilmore, would love to have input, particularly to the point where we could actually share code
20:42:30 <dgilmore> walters: well it would need to be integrated into pungi
20:43:13 <dgilmore> pungi would need to know how to make the atomic tree and make the installer bits and then wrap it all up
20:43:53 <dgilmore> walters: we use kickstarts as the input to everything
20:44:28 <walters> ok, i'll look at pungi
20:44:43 <walters> if you can look at the script in more detail that'd be great, and we can sync up in the next meeting
20:44:47 * stickster finds https://git.fedorahosted.org/git/pungi.git
20:45:40 <stickster> #action dgilmore examine ISO creation script in https://lists.projectatomic.io/projectatomic-archives/atomic/2014-July/msg00002.html
20:45:56 <stickster> #action walters check out pungi at https://git.fedorahosted.org/git/pungi.git
20:46:16 <stickster> #agreed sync at next meeting to see how ISO creation could be handled in pungi
20:46:39 <stickster> Anything else before we close?
20:47:15 <stickster> OK, closing in 10
20:47:22 <stickster> 5
20:47:33 <stickster> Thanks everyone!
20:47:35 <stickster> #endmeeting