13:00:55 <gwd> #startmeeting Virt SIG 13:00:55 <centbot> Meeting started Tue Apr 21 13:00:55 2015 UTC. The chair is gwd. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:55 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic. 13:01:02 <gwd> #chair lars_kurth 13:01:02 <centbot> Current chairs: gwd lars_kurth 13:01:06 <gwd> #chair kbsingh 13:01:06 <centbot> Current chairs: gwd kbsingh lars_kurth 13:01:08 <gwd> #chair lsm5 13:01:08 <centbot> Current chairs: gwd kbsingh lars_kurth lsm5 13:01:15 * sbonazzo here 13:01:21 <gwd> #chair sbonazzo 13:01:21 <centbot> Current chairs: gwd kbsingh lars_kurth lsm5 sbonazzo 13:01:54 <gwd> kbsingh can you give a quick one-liner of what you wanted to talk about, so we can schedule it in the minutes? :-) 13:02:44 <kbsingh> ok 13:03:14 <kbsingh> there is a lot of container and ecosystem around container work shaping up and coming at the centos side of things - so far, the way things are setup, we are split in that into 4 groups 13:03:49 <kbsingh> the vir-sig owns the container tooling, the atomic and core own container hosting, and instance-sig has the base container space itself, with the layered and orchestration stuff largely on the lose 13:04:10 <kbsingh> is it worth scoping up the change, re-org'ing a bit, rethinking how we do the container specific stuff in CentOS 13:04:14 <kbsingh> see ? one liner :) 13:04:26 <lsm5> kbsingh: SGTM 13:04:31 <gwd> Nice. :-) 13:04:34 * dcaro here 13:04:40 <gwd> #chair dcaro 13:04:40 <centbot> Current chairs: dcaro gwd kbsingh lars_kurth lsm5 sbonazzo 13:04:47 <gwd> Well shall we start with Docker update? 13:05:04 <kbsingh> also, not looking for a decision today but just to get this idea up in the air so we can all start thinking about things, we dont have a problem as such to solve right now 13:05:07 <kbsingh> yeah, lets docker 13:05:10 <lsm5> gwd: 1.6.0-1 in virt7-testing as of yesterday, this build only tracks upstream sources 13:05:14 <gwd> #topic Docker update 13:05:29 <lsm5> gwd: i'm open to creating new builds with different names to make anyone else happy 13:05:58 <lsm5> gwd: but I'd like to keep 'docker' in virt7 as something that tracks upstream as a fallback for people unhappy with docker in Extras 13:06:30 <lsm5> coz both docker upstream and many centos users need a vanilla docker somewhere on centos 13:06:44 <lsm5> gwd: that's about it from the docker side 13:06:57 <kbsingh> if i may, just one point 13:07:01 <gwd> lsm5: So the docker in Extras has the RHEL patches...? 13:07:14 <kbsingh> we have too many dockers, and we dont have a user story around any ( except the one in CentOS Extras ) 13:07:32 <lsm5> gwd: yes, as the one in extras is inherited from RHEL itself :) 13:07:42 <gwd> Ah, right. 13:07:47 <kbsingh> if we can, we should try and drive the user story a bit - having a nightly docker build is awesome, as is kubernetes nightly, but thats not really meant for the end user as such 13:08:20 <lsm5> kbsingh: could we say nightly builds coz we can find out about problems earlier? 13:08:36 <lars_kurth> kbsingh: do you have any concrete ideas on how to drive the user story 13:09:11 <gwd> Well it sounds like 1.6.0-1 is just an up-to-date release, right? 13:09:41 <lsm5> gwd: yup, latest upstream release 13:09:44 <gwd> I think we want the main docker in the Virt SIG to be a "product"; and I think it makes sense to just take one and stick with it. 13:09:54 <kbsingh> lars_kurth: i think if the tag's and repo targets line up, we can then use the 'unstable' or 'testing' or 'nightly' or 'devel' or something for those, and a 'released' for the regular builds 13:10:03 <gwd> So I think lsm5 you should just try to make the best decision fro what you think the "product" should look like. 13:10:35 <gwd> If it seems like most people want a vanilla Docker, I think that's perfectly sensible. 13:10:36 <lsm5> gwd: sure thing 13:10:48 <lars_kurth> kbsingh: as far as I recall that's what we wanted to get to. We just have to get through the release cycle once, with signing and all. Then we can do this in future again when a new release comes up 13:12:21 <gwd> OK, Xen? 13:12:34 <gwd> #topic Xen update 13:12:47 <gwd> #info Build C6 kernel and Xen packages on C7, in virt-testing 13:13:10 <gwd> #info Should be testing the new virt-xen-44 tags later this week 13:13:16 <gwd> That's about it from me. 13:14:00 <gwd> #topic oVirt update 13:14:11 <sbonazzo> no much info from ovirt side on virt sig 13:14:18 <sbonazzo> we're finalizing 3.5.2 upstream 13:14:56 <sbonazzo> and if koji tags are there we may try to release some 3.5.2 GA packages there 13:15:17 <gwd> Cool. 13:15:26 <sbonazzo> looking at https://bugs.centos.org/view.php?id=8407 seems everything is in place thre 13:15:28 <sbonazzo> there 13:15:38 <sbonazzo> not verified yet 13:16:02 <sbonazzo> I think that's all on oVirt. dcaro ? 13:16:13 <lars_kurth> Do we have anyone being blocked/stalled on anything? 13:16:15 <dcaro> sbonazzo: nothing on my side 13:16:33 <gwd> OK, kbsingh -- do you want to talk about container space? 13:16:50 <gwd> #topic Where containers should live 13:18:10 <kbsingh> there is a bunch of stuff happening, and coming up 13:18:36 <kbsingh> in some cases, its also a case of me working with other upstreams to adopt and have them build their stuff / devel and deliver across CentOS 13:19:07 <kbsingh> there has been talk of the public hosted CentOS Container Pipeline that allows anyone to throw a definition ( docker or otherwise ) and have it rendered, tested, delivered, via centos infra etc 13:19:34 <kbsingh> and there are artifacts that fall out, eg: vagrant tooling, cloud instances, test harness's etc - even hosted instances of the same code 13:19:53 <kbsingh> then there is the delivery mechanisms, eg: hosting a docker registry at mirror.centos.org for content in and around the distro etc 13:20:17 <kbsingh> so the question really is, is that space becoming large enough to merit its own group - so as to not drown up the other work in the virt sig 13:20:37 <kbsingh> and then to rope in the other stuff happening in the instance sig ( like the docker images and the gce container stuff etc ) 13:22:15 <lars_kurth> kbsingh: is there some concrete public proposal already somewhere, or are you exploring at this stage? 13:24:28 <gwd> kbsingh: So most of that sounds kind of out of the original remit of the Virt SIG, which was mainly to deliver upstream virtualization technologies, right? 13:24:56 <lars_kurth> lsm5, gwd, ...: what are your thoughts on the above? 13:25:53 <kbsingh> lars_kurth: there is stuff happening already 13:26:56 <kbsingh> lars_kurth: we have an instance of the Openshift builder service online at ci.centos.org as of yesterday, thats a layered image build service to extend the community rpm space in cbs.centos.org 13:27:05 <lsm5> lars_kurth: gwd perhaps one big place for the products docker/atomic(the tool)/registry and anything that use these tools might be the way to go 13:27:53 <lars_kurth> I don't have a strong view either way. 13:28:32 <kbsingh> lets think it though 13:28:39 <gwd> kbsingh: I think part of the problem is you're speaking too concisely. :-) What's this Container Pipeline thing about? 13:28:43 <kbsingh> maybe in a couple of weeks when we meet again, we can try to quantify what that means 13:29:28 <lars_kurth> kbsingh: agreed - also if I was you, I would evaluate against the original SIG model. There are really a few ways of how to look at it 13:29:31 <kbsingh> gwd: good question :) I will try and put it into an email 13:29:54 <gwd> Who do you expect to throw a definition to this system, what kind of image will it build, and who might want to use it? 13:30:15 <lars_kurth> To some extent the virt SIG is a collection of similar technologies (almost like a "horizontal" market segment) 13:31:01 <lars_kurth> Putting together a SIG that represents a complete stack a la docker/atomic/.../registry is more like a vertical and that may be more user friendly 13:31:02 <gwd> lars_kurth: But depending on how it's set up, you could imagine being able to test and post VM images as well. 13:32:07 <lars_kurth> Of course the two models are not really in conflict with each other: you can package a vertical from t 13:32:10 <lars_kurth> gwd: agreed 13:32:42 <lars_kurth> oops: ... a vertical from different SIGs 13:33:21 <gwd> But as kbsingh said, the amount of people coming in and work going on for such a system would be far more than what's going on in the Virt SIG today. 13:33:28 <kbsingh> i need to rebase over from here guys, will catchup later. and will do details for the pipeline in an email 13:33:53 <gwd> kbsingh: OK, cool. Look forward to it. 13:33:53 <lars_kurth> OK: maybe a public email discussion before the next meeting is the way to go 13:34:08 <gwd> #topic AOB 13:34:21 <sbonazzo> no OB from me 13:34:28 <lars_kurth> GSoC? 13:34:38 <gwd> One more bit of Xen update -- there have been a number of good submissions for the GSoC "Xen in a Box" proposal. 13:35:16 <gwd> I think I'm not supposed to say anything concrete publicly, but things look pretty good for getting a student for that project. 13:36:27 <gwd> Anything else? 13:36:29 <lars_kurth> gwd: as far as I can see the publication date for accepted students ins on the 27th 13:36:39 <lars_kurth> Not from me 13:37:15 <gwd> OK, great -- thanks everybody! 13:37:18 <gwd> #endmeetingg 13:37:23 <gwd> #endmeeting