18:02:06 #startmeeting 18:02:06 Meeting started Tue Sep 23 18:02:06 2014 UTC. The chair is jzb. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:06 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:02:06 Meeting started Tue Sep 23 18:01:31 2014 UTC. The chair is jzb. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:06 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:02:19 * stickster expects dgilmore since we moved the time for his benefit 18:02:48 I am here but I need to leave in 5 minutes 18:02:52 I have a one off thing 18:03:20 ok, then 18:03:31 walters: do you have anythign for dgilmore he can knock out in 5 minutes? 18:03:39 im still dealing witha bunch of release things 18:03:52 ok, shall we reschedule? 18:04:10 im trying to get permission to write to /pub/fedora/linux/ so i can put the atomic tree there 18:04:20 the main thing i have from the projectatomic dev side is continued work on compose scripts 18:04:21 * oddshocks here 18:04:28 though I guess we should maybe put it in stage 18:04:37 im really not sure where exactly is best 18:04:43 geppetto, did you have a link to any packaging work? it'd be good to run that by digilmore to see if we can converge on running it in fedora infra 18:05:17 walters: And for the latter packaging work, a nod from nirik would be helpful, although he's out this week IIRC 18:05:37 other than that, where did we leave the metalink/mirrorlist bits? was that waiting on having bits in stage? 18:06:23 * stickster really wishes people would spell out the status a little better. There are too many "bits" around to understand what precisely is going on 18:06:51 stickster: I honestly do not know what is what 18:06:52 No slight intended 18:07:13 stickster, compose script is here: https://git.fedorahosted.org/cgit/fedora-atomic.git/tree/compose this is presently distinct from the fedora-infra one 18:07:18 I do know that the atomic tree is not public yet because i can't write to where I want to put it 18:07:41 who owns that? infra? do we need to file a ticket? 18:07:58 walters: Thanks, I was going to ask that :-) 18:08:21 * stickster got off phone -- jzb, sorry for not being available on the hour 18:08:27 stickster: no worries 18:08:33 * jzb passes baton to stickster 18:08:35 walters: releng 18:08:39 #chair stickster 18:08:39 Current chairs: jzb stickster 18:08:39 Current chairs: jzb stickster 18:08:48 walters: in fedora it actually has almost nothing to do with infra 18:08:57 #chair walters oddshocks dgilmore 18:08:57 Current chairs: dgilmore jzb oddshocks stickster walters 18:08:57 Current chairs: dgilmore jzb oddshocks stickster walters 18:09:05 * dgilmore needs to run 18:09:17 dgilmore: Can you please come back to us with a time you can meet at more length? 18:09:39 dgilmore: That way we can have a detailed discussion with everybody at the same time :-) 18:09:42 stickster: this time is normally fine 18:09:53 stickster: i just have something on today 18:09:56 dgilmore: walters: Can everyone do tomorrow this time? 18:10:08 stickster: im taking PTO for the rest of teh week 18:10:20 and maybe monday and tuesday next week 18:10:34 stickster: ive had 1 day off all year 18:10:39 I kinda need a break 18:11:02 dgilmore: I understand. :-) Is this something pbrobinson can help with while you're gone? 18:11:15 stickster: sadly not yet 18:11:44 stickster: my team already has a meeting setup with walters 18:11:51 dgilmore: Ah, that's helpful 18:12:34 Sorry to belabor this. It's hard from my POV to understand who's got the ball without a longer discussion. 18:12:39 stickster: but none of this really has anything much to do with infra 18:12:44 other than the mm side 18:12:44 * jbrooks here 18:12:46 er rather [ball1, ball2, ball3...] 18:13:13 anyway i really need to run, sorry guys 18:13:40 dgilmore: OK, we'll plan to see you next week. In the meantime... 18:14:03 walters: Can we come back to the other two items for a moment? 18:14:24 sure 18:14:33 #info dgilmore is short on time today due to F21 release tasks leftover, will have to defer to next week 18:16:17 #topic Compose scripts 18:16:40 walters: So wrt the compose scripts, what is the next thing that needs to happen? 18:16:50 for fedora 21, the deliverable is just a cloud image 18:17:01 for fedora 22, we want at least an installer ISO 18:17:17 and likely other vehicles such as boot-to-LIVE 18:17:33 the compose script there does installer iso plus cloud, plus vagrant 18:17:49 we're trying to cover both the local dev case and ideally something for production 18:18:06 walters: So we have e.g. http://alt.fedoraproject.org/pub/alt/stage/current/Cloud/Images/x86_64/ right now 18:18:13 an issue with the fedora mainline scripts is (AFAIK) there's no orientation or facility for running them localy 18:18:24 which is mostly fine because the underlying tools are the same 18:18:45 walters: So someone can't replicate the compose locally? 18:19:10 the tools are all there, it's a question of scripts/configuration 18:19:38 walters: the packaging stuff I did is in the upstream git repo. 18:20:08 walters: It's not perfect, name/file locations/etc., but it turns the code into a package … so there's that :) 18:20:16 #info Fedora mainline compose scripts don't have the capability to be run locally; tools are the same, but configuration may not allow someone to actually make images 18:20:27 geppetto, ah cool, i see 18:20:38 fudge sorry. here 18:20:39 anyways this is something to talk about when dgilmore is back 18:21:16 * geppetto nods 18:21:42 walters: OK, so correct me if I'm wrong: we need to update the compose scripts so they can be run both by Fedora releng and by other contributors? 18:22:34 it's a goal 18:23:12 walters: Is that the most proximate goal at this point? 18:23:17 this is much more important now that Fedora is shipping content types other than the same RPM repositories that it's done since the start 18:23:35 s/other than/in addition to/ 18:23:59 *nod 18:24:15 well the highest priority goal is a cloud image that boots in at least KVM and OpenStack and an ostree repo with a mirrorlist that we can pull updates from 18:25:27 * stickster copies for notes 18:25:43 #info highest priority goal is a cloud image that boots in at least KVM and OpenStack and an ostree repo with a mirrorlist that we can pull updates from 18:27:32 OK, so it seems like we need something analgous to pungify that does the cloud stuff needed in build-branched and build-rawhide 18:27:40 but we'll need releng to confirm that 18:27:56 * stickster feels like we are spinning wheels since last meeting on this 18:28:39 But since I'm not deeply educated on how either the atomic and releng stuff work, I could be wrong 18:29:11 Oh, I see -- there is a build-cloud-images now 18:29:39 walters: I don't want to waste everyone's time while I'm playing 20 Q's here :-) 18:29:57 geppetto: smooge may be able to help look at your packages and give some advice 18:30:10 smooge: Does that sound like something you could do? 18:30:19 stickster: sounds fine to me 18:30:20 OK, sorry 18:30:28 you now have my full attention, whatever that's worth 18:30:39 jzb: You're just in time to see the tail end of my being ignorant 18:30:56 stickster, I am trying to still catch up with conversation I missed. 18:32:08 isn't geppetto the RPM maintainer? I would feel weird giving advice to the guy who writes RPM :) 18:32:19 but sure.. 18:32:34 stickster, right. The compose script discussion can wait a week or two I'd say - we can just keep running composes on atomic01.qa too 18:32:53 smooge: I dunno, geppetto sounded like he wasn't sure whether e.g. file locations are in the right place. Knowing everything about rpm/yum doesn't make one a Fedora packaging expert for things that have to go in infrastructure repo, necessarily 18:33:23 smooge: At the same time, just trying to connect up dots so as to reduce "not-sure-ness" 18:33:54 walters: Is there anything that needs done in the intervening week that would help? 18:33:56 ok I can talk with geppetto after this meeting to see where he needs help and see if I am the guy who can help it 18:34:04 smooge: Thanks 18:34:49 smooge: cool 18:35:48 OK, any thing else we need to cover? 18:35:55 * stickster hands baton back to jzb :-) 18:36:01 are we going to try to reschedule with dgilmore or just hold out for next week? 18:36:20 if there's someone else that can pick up the thread of helping to mirror the repo content with a metalink that'd be awesome 18:36:21 well next week would be when dgilmore is around 18:36:37 FWIW I am putting the final touches on the x86_64 atomic images for Fedora Cloud 21 RC1, which will be listed here when I'm done: https://oddshocks.fedorapeople.org/ 18:36:43 right now there's just the x86_64 base AMIs 18:36:49 oddshocks, oh cool! 18:37:14 there will also be i386 base images but I have to finish a thing or two with those, as well. that's what i'm working on right now 18:38:28 had to fix a couple things with the atomic images because they require HVM, and HVM requires a few changes to the standard configurations I was using for the base images. so patching fedimg appropriately 18:38:41 walters: that still requires MM changes, yes? 18:38:52 jzb, i believe so 18:39:08 walters: OK, I thought someone was working on that. 18:39:24 #action jzb figure out who's got the ball with mirrormanager. 18:39:44 walters: jzb: that's something I might be able to help with. it's on my list, I just wasn't sure where to start, and have been focusing on these RC1 images 18:40:24 oddshocks: OK. Can you maybe take the time to look at it this week and at least give an estimate whether it's something you can do? 18:40:36 and a rough guess when it would be done, what you need, etc.? 18:41:03 jzb: sure, are there any existing resources I can use to get an idea of what sort of code needs to be added to MM? 18:41:28 walters: do you have any feedback on that ^^ 18:41:30 ? 18:42:32 oddshocks, basically, ostree --repo=repo summary -u will generate repo/summary 18:42:38 that file needs to be the target of the metalink xml 18:43:06 is this one of those things that is actually harder to do than it sounds? :P 18:43:43 mmm... 18:44:07 oddshocks: no, it's totally easy. You should totally commit to it. ;-) 18:44:12 haha 18:44:19 either way put me down for checking it out 18:44:26 i'll keep people posted 18:45:28 oddshocks: BTW -- thank you for work on RC1 and Alpha (pre-released!) images 18:45:48 #action oddshocks will look into Mirrormanager changes and report back 18:45:58 #info jzb thanks oddshocks profusely 18:46:10 OK, I think that's all we've got today 18:46:33 unless someone has a major last-minute item, I'll go ahead and close the meeting 18:47:12 oddshocks, where will you send status about the RC1 AMI? fedora-cloud-list ? 18:48:32 stickster: no prob! 18:48:52 walters: want me to send out an email when all 3 sets of RC1 images are out? 18:48:58 32 and 64 base and 64 atomic, that is 18:49:09 oddshocks, yes please! 18:49:17 * oddshocks notes as such 18:50:32 #action oddshocks to send note to cloud@lists.fedoraproject.org when AMIs are ready 18:51:24 OK, calling it for today 18:51:28 thanks everybody 18:51:34 #endmeeting