14:08:17 #startmeeting 14:08:17 Meeting started Thu Apr 24 14:08:17 2014 UTC. The chair is jzb. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:08:17 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:08:26 #topic roll call 14:08:30 * roshi is here 14:08:37 .fasinfo jzb 14:08:38 jzb: User: jzb, Name: Joe Brockmeier, email: jzb@zonker.net, Creation: 2010-04-20, IRC Nick: jzb, Timezone: America/Chicago, Locale: en, GPG key ID: A0207CD4, Status: active 14:08:40 jzb: Approved Groups: marketing cla_done cla_fpca cla_fedora 14:08:46 * geppetto is here 14:08:59 * number80 ahoj 14:09:02 * red_trela 14:10:00 OK, let's look at the agenda 14:10:06 #topic Meeting tickets 14:10:12 https://fedorahosted.org/cloud/query?keywords=~meeting&status=!closed 14:11:02 Let's start with #16 14:11:14 #topic rename cloud spin kickstart to distinguish the cloud base image 14:11:48 #info https://fedorahosted.org/cloud/ticket/16 14:12:02 discussion on that one? 14:12:10 +1 on the proposal 14:12:26 I'm +1 14:12:41 +1 14:12:43 I'm assuming mattdm is +1 since it's his idea 14:12:45 +1 if I can vote on it 14:12:53 +1 14:13:29 +1 14:13:31 :) 14:13:48 OK, I think we have that one settled. 14:13:59 mattdm: I thought you weren't going to be here :) 14:14:04 is it necessary to have cloud-base-base.ks and cloud-base.ks, though? i.e. what's going to be the difference? 14:16:14 anybody? 14:16:52 I'm not familiar enough with the image creation process :-/ 14:17:18 red_trela: I don't think we have the folks in attendance to discuss. Can we start a thread and try to solve this on the list? 14:18:01 if we only just agree on the image name, yes. the second part of the ticket is the ks naming, though so we shouldn't decide that yet. 14:18:32 OK 14:18:37 Moving on... 14:19:24 #topic File F22 change: Re-factor cloud-init 14:19:31 this is ticket #23 14:20:08 though, reading through the ticket, this may be deferred 14:20:36 #info https://fedorahosted.org/cloud/ticket/23 14:21:26 we need someone to step up for refactoring cloud-init, provisionally 14:22:11 anybody? 14:22:20 * jzb feels like he's talking to himself here. 14:22:21 I'm more in favor to improve min-metadata service 14:23:01 number80: you want to update the ticket? 14:23:14 I'm in favor for any of the alternatives, over the legacy cloud-init 14:23:44 but if we stick with legacy, it must be refactored some 14:23:44 nope, because i'm not sure that it will be acceptable for open{stack,nebula} folks for instance 14:23:51 red_trela: we're not concerned that we'll lose / fail to attract people who expect cloud-init? 14:24:43 jzb: that sure is a big question 14:25:18 jzb: do we have access to cloud instances ? 14:25:26 jzb: since we want to be "First", we might be ready to take that route 14:25:31 red_trela: my general observation is that it's one of those "I expect this, and even if there are good reasons not to have it, I'm just going to pass it by b/c it's not hitting one of my checkboxes." 14:25:47 jzb: also, CoreOS is hugely popular and develops their own cloud-init 14:25:50 I could work on implementing some missing features, but it would be expensive if I did on my own EC2 account 14:26:08 https://github.com/coreos/coreos-cloudinit 14:26:14 number80: I don't think we have an account as a group, no. 14:26:21 number80: that's not to say it's something we couldn't come up with 14:26:28 number80: define "expensive" 14:26:52 jzb: I was thinking about eucalyptus or openstack clouds hosted in fedora-infra 14:26:53 number80: open a ticket with fedora-infrastructure to get an account on their cloud 14:27:11 number80: i.e. the fedora infrastructure openstack cloud 14:27:20 red_trela: great 14:27:22 https://fedorahosted.org/fedora-infrastructure/ 14:27:34 ^^ unnecessary googling, probably :-) 14:27:40 :) 14:28:02 red_trela: CoreOS is getting a lot of attention, I'm not sure about "hugely popular" 14:28:14 not sure how much, if any, real world adoption is going on there. 14:28:40 jzb: not sure how much, if any, real world adoption we are getting at the moment... 14:28:45 (in the cloud) 14:28:48 red_trela: there's that 14:29:04 OK, anything else on this ticket? 14:29:11 * roshi has nothing 14:29:26 I take it we kind of need to leave this one for further discussion. 14:29:34 I'll add myself as CC, but I need to get more visibility 14:29:55 number80: thanks 14:30:05 number80: can I put you down as taking lead on that? 14:30:13 I'm sure mattdm would appreciate it :-) 14:30:14 yes 14:30:45 number80: could you start a thread on the ml to discuss the various cloud-init alternatives? 14:30:52 #action number80 to take lead on Ticket #23 (https://fedorahosted.org/cloud/ticket/23) 14:31:28 #topic Ticket #45 need three packages added to the fedora21 cloud image 14:31:55 sdake has requested that we add os-collect-config, os-apply-config, and os-refresh-config to the Fedora Cloud images. 14:32:59 reading the thread, I'm not entirely sure this is needed - 14:33:04 any thoughts, comments, etc.? 14:33:24 have we established that these don't need to be in the image by default, and can be added later, or do we need to look at adding them? 14:33:26 it looks like it's requested for one thing 14:33:41 I won't vote since i might be suspected of being partial here 14:33:41 but wasn;'t a long term goal getting the python dep out of the image? 14:34:05 with the previous cloud-init ticket, yeh 14:34:13 now we add a second python :-o 14:34:15 at first it seemed he wanted the packages included for tripleO but as I raised concerns he dropped that use case and went with Heat instead. So right now we have heat-cfntools (109k according to yum info) and the new tools use 50MB (installed size) 14:35:12 that really goes against our efforts of minimizing the footprint and sdake has failed to come up with a (good) reason for it so far. 14:35:14 Does anyone know how possible it would be to have it use py2 until everything else is using py3? 14:35:31 roshi: only our of the docker image, which would not ship those tools 14:35:49 s/our/out/ 14:36:00 geppetto: though I'm a pythonista, I'd look forward dropping interpreters in the base image 14:36:05 roshi: the other images ship yum, so they need python ;) 14:36:06 depends on the lib and it's deps - py2 vs 3 is a giant PITA in my experience 14:36:24 I'd also point out 14:36:24 number80: right … and I agree, but 1 python is still better than 2 :) 14:36:32 that this is only aimed at one IaaS 14:36:38 ah - that makes sense red_trela - finally getting my head wrapped around it 14:36:40 AFAICT 14:36:57 it only benefits use cases with Fedora on OpenStack, not Euca, AWS, GCE, or CloudStack 14:37:04 (or am I wrong on that?) 14:37:16 jzb: right 14:37:28 geppetto: sure 14:37:37 so we'd be looking at adding another runtime, and embiggening the image just for that use case. 14:37:38 jzb: and only when using Heat (i.e. Orchestration) 14:37:47 embiggening 14:37:55 * roshi is going to use that word going forward 14:38:00 jzb: heat could be used with other cloud infrastructures too 14:38:06 I think that we'd either want to look at having a separate spin for that or providing a Puppet recipe, etc. instead. 14:38:10 (I once used it with AWS) 14:38:18 number80: ah, OK - would people do that? 14:38:26 number80: that's a serious edge case :) 14:38:34 I think Heat is actually implementing a native AWS tool, right? 14:38:53 CloudFormation, yes 14:38:59 jzb: not sure but Heat has some compatibility with CloudFormation (its AWS equivalent) 14:39:21 OK 14:39:22 but I think they're moving away from compatibility 14:39:31 hi folk 14:39:35 sorry i am late 14:39:37 long story 14:40:16 hi frankieonuonga 14:40:25 let's put it to a vote here 14:40:27 jzb: hi mate... 14:40:39 can someone kindly get me up to speed on DM 14:40:39 who's in favor of adding these packages? 14:40:58 -1 unless sdake can come up with new good arguments 14:41:10 -1 14:41:39 frankieonuonga: we're voting on this ticket https://fedorahosted.org/cloud/ticket/45 14:41:50 thanks 14:42:13 -1 unless we get further info that provides a better rationale for adding them. 14:43:15 number80? frankieonuonga? 14:43:33 geppetto: ? 14:43:42 I am voting agaist it cause there is not enough to to back it up 14:43:50 jzb: I'm abstaining to vote because it involves sdake and I won't be considered impartial 14:43:51 one really needs to give a good argument 14:44:05 here i feel at this moment it is weak 14:45:02 summoning mattdm 14:45:31 * mattdm reads back 14:46:25 jzb: I'm not sure … I guess -1 for including it in images for everyone (although that vote doesn't really matter :) … would be nice to get some more info. though, and why they can't just have their own image 14:46:58 I had some further conversation with sdake but can't remember it all :) 14:47:17 I think we're currently at "needs more info, but probably not." 14:47:28 I can be +1 to that statement 14:47:37 which I guess joins everyone else's -1s 14:48:33 and we can maybe look at adding an openstack-optimized spin? I'm not exciting about going down that road, but when stuff starts getting loaded up with environment-specific agents it might be the best way. 14:48:35 OK, I'm reassigning this one to myself to track 14:49:07 mattdm: yeah 14:49:27 #action jzb to follow up on Ticket 45, will herd cats until we come to a conclusion on it. 14:50:03 #topic Project Atomic and Fedora Docker Host Image 14:50:03 jzb: I was actually just going to write it since I was already the in the discussion on the ticket 14:50:17 \o/ 14:50:19 red_trela: already done :-) 14:50:41 https://fedorahosted.org/cloud/ticket/46 14:50:41 jzb: okay ;) 14:50:53 so questions 14:51:02 I'm so +10000 for that 14:51:11 1) do we want the Fedora Docker host to be the Atomic host 14:51:16 mattdm is heavily +1 on an All-in-on-Atomic image 14:51:20 apparently number80 likes it 14:51:35 it should be pretty obvious where I am on that, but +1 14:51:53 I'm already annoying everyone with how awesome is atomic :) 14:51:53 +1 14:52:05 * roshi doesn't vote because he only has a passing understanding of Docker/Atomic 14:52:28 Why didn't Colin open this? 14:52:54 Not sure 14:53:38 atomic is grown bigger than just colin's baby :) 14:53:46 Anyway … do you have a plan other than to have the docker host be ostree/Atomic based? 14:53:59 so, what do we understand by this. Basically cp project-atomic.qcow2 fedora-21-atomic.qcow2? 14:54:18 geppetto: define "other plan"? 14:54:35 geppetto: Atomic also pulls in geard, cockpit, etc. 14:55:03 I want to get this under the fedora umbrella, _rather_ than the 'cp' approach 14:55:14 +1 with mattdm 14:55:43 jzb: I'm just not sure what the point of voting is … it seems like asking should we have the linux kernel in our docker host … AFAIK all the plans for a docker host are using atomic, so we'd have to be pretty crazy to not do so. 14:55:51 besides, the atomic host could also be the base for the openshift image 14:56:04 (not the _whole_ thing under fedora, of course, but I want to make sure that the part that _is_ under fedora is official and integrated) 14:56:13 geppetto: we could have a non-Atomic docker host 14:56:43 I don't see that as an advantage, but I suppose it'd be possible someone would be against the rpm-ostree stuff, for example. 14:56:45 geppetto let's vote for "we're not crazy", then :) 14:57:01 mattdm: why do you keep excluding me? 14:57:09 jzb lol 14:57:20 mattdm: Yeh, I mean … for official counts, put me down on the not crazy side :) 14:57:33 OK, I ... think we actually have majority there. Yay 14:57:39 mattdm: okay, let me ask that differently. is the question whether to use rpm-ostree or is it whether to do all the project atomic stuff (rpm-ostree, geard, cockpit, etc.) and ignore the alternatives (accepting the guys at project atomic choose the best stuff) 14:58:45 red_trela I think it is to use the project atomic patterns as a foundation (including all that stuff), and possibly do whatever else fits within that and makes sense 14:59:41 won't it be possible to get projectatomic.io as a co-owner of this feature ? 14:59:54 number80 that would be cool, yes. 15:00:17 for example, jzb :) 15:00:44 I will do what I can, and push for other folks 15:00:47 :) 15:01:09 moarr collaboration 15:01:28 so, do we want to open new tickets for specific items and consider this one a yes? 15:01:32 or just use this as the tracker? 15:01:49 * mattdm has a lot of informal "yeah, I'd love to help in fedora" comments that we _should_ be able to turn into commitments. :) 15:02:08 jzb: tracker 15:02:22 sigh, I didn't have half the time I'd liked to dig into project atomic so I feel I'm making a somewhat unfounded decision...but throwing in a +1 as well 15:02:41 always having a hard time to resist the leading edge ;) 15:03:00 red_trela: that makes two of us 15:03:00 it's chronic :) 15:03:01 red_trela: time seems to be our enemy 15:03:08 * mattdm is glad to have fedora near the leading edge in the cloud area for once :) 15:03:23 or, more accurately, having to eat, sleep, pay bills, and scoop litter boxes. 15:03:43 OK 15:03:45 eating is that thing I always seem to forget to do during the workday 15:04:03 roshi: yeah... that's definitely not my problem. Forgetting to eat would be a feature. 15:04:24 sleep is the thing I would like to go do since our 1h is over...so can we please wrap this and try having more regular meetings from now on? :) 15:04:40 I have one thing 15:04:44 red_trela: I like the way you think 15:04:53 discussion can happen on list or wherever 15:04:54 do we need to address any of the other tickets urgently today? 15:05:02 https://fedorahosted.org/cloud/ticket/41 15:05:03 or shall we table them for next week? 15:05:20 anyone here want to look at the gce effort? 15:05:21 I created some draft stuff for testing I'd like feedback on :) 15:05:53 mattdm: erjohnso spoke up about that at the beginning of the meeting. 15:06:00 roshi: they are fairly high up on my todo list :) 15:06:03 yup, i'm here to assist :) 15:06:12 erjohnso is from google and wants someone here to connect with 15:06:30 if you give me a pointer to your standard image creation tools, I can give them a try 15:06:35 roshi: thanks for that, can you send a note to the list? 15:06:40 yeah 15:06:59 #action roshi to send note to the list for review of his QA draft 15:07:01 erjohnso: is it possible to get credentials for testing purposes ? 15:07:05 roshi: thanks again! 15:07:06 I would like to help but I can not promise to touch this until next week tuesday 15:07:15 then my schedule is free again 15:07:27 if I felt more qualified to call myself a cloud person I would volunteer mattdm 15:07:31 erjohnso: if you need any help just ping me 15:07:45 and since we brought up GCE, has anyone an update on packaging GCE utilities progression ? 15:07:53 roshi: Only one way to learn, through yourself into the dip end 15:08:17 true enough, I suppose 15:08:18 I can likely get a free project created and give folks access to it - I'll need to check on that though 15:08:24 frankieonuonga: deep end? 15:08:30 frankieonuonga: great! 15:08:49 erjohnso the tool that we're migrating to is imagefactory, fed with this kickstart file https://git.fedorahosted.org/cgit/spin-kickstarts.git/tree/fedora-x86_64-cloud.ks 15:08:53 jzb: yeah...like just get yourself into the mix of things and find your way around :-) 15:08:56 erjohnso: I may contact you after you'll send a note to the list :) 15:09:03 zooz (Vaidas Jablonskis) has done some work on this 15:09:12 I'll try to dig up the mailing list posts 15:09:49 thanks 15:09:53 figure fedora legal would want to talk with google about such accounts just as they do with hp cloud... 15:10:00 much appreciated. sorry, new to the fedora community (old school redhat user though) - what list should I post to? 15:10:08 red_trela that is still in the black box state 15:10:31 red_trela: I thought that samkottler has sorted out this issue with legal 15:10:51 erjohnso https://lists.fedoraproject.org/pipermail/cloud/ 15:11:06 mattdm: thx 15:11:14 number80: not yet afaik. and it's taking forever already so don't expect to use gce like that soon :/ 15:11:41 :( 15:11:59 remember when legal (or actual laws, even) had no idea about IT and everything was just great? ;) 15:12:40 I'm not bound to their whims :> 15:12:58 so, who got the wrapping paper for this one? 15:13:32 red_trela: is that a call for adjourning? 15:13:49 jzb: please :) 15:13:55 +1 15:13:58 +1 15:14:01 +1 15:14:03 +1 15:14:03 it's already Friday over here ;) 15:14:19 I'm pretty sure I can count mattdm as +1 too 15:14:23 so, meeting adjourned 15:14:27 #endmeeting