17:02:09 #startmeeting cloud_wg 17:02:09 Meeting started Wed Dec 9 17:02:09 2015 UTC. The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:02:09 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:02:09 The meeting name has been set to 'cloud_wg' 17:02:28 #topic roll call 17:02:32 .hellomynameis kushal 17:02:33 kushal: kushal 'Kushal Das' 17:02:40 .hello jasonbrooks 17:02:41 jbrooks: jasonbrooks 'Jason Brooks' 17:03:11 .hellomynameis dustymabe 17:03:11 .hello sayanchowdhury 17:03:11 dustymabe: dustymabe 'Dusty Mabe' 17:03:15 sayan: sayanchowdhury 'Sayan Chowdhury' 17:03:31 #chair kushal sayan jbrooks jzb 17:03:31 Current chairs: dustymabe jbrooks jzb kushal sayan 17:03:35 .hello tflink 17:03:36 tflink: tflink 'Tim Flink' 17:03:38 .hellomynameis jzb 17:03:39 jzb: jzb 'Joe Brockmeier' 17:03:57 * tflink may have to disappear during the meeting, though 17:03:58 #chair tflink 17:03:58 Current chairs: dustymabe jbrooks jzb kushal sayan tflink 17:04:30 dustymabe, please push https://fedorahosted.org/cloud/ticket/143 at the top while discussing the tickets :) 17:04:35 we haven't seen adimania in a while 17:05:04 anybody know if he is still around? 17:05:04 kushal: you read my mind 17:05:05 .hello maxamillion 17:05:06 maxamillion: maxamillion 'Adam Miller' 17:05:13 sorry I'm late 17:05:24 kushal: I was going to ask about that, and whether folks were tracking other changes that might affect Atomic/Cloud 17:05:29 jzb, sorry for not writing the mail before. 17:05:34 kushal: no worries 17:05:50 ok we'll start with that then 17:05:56 #topic Proposals for F24 features 17:06:03 #link https://fedorahosted.org/cloud/ticket/143 17:06:11 dustymabe, action items from last meeting? 17:06:13 :) 17:06:15 We can do that after this :) 17:06:17 ahh sigh. 17:06:20 ok 17:06:33 So Fedora 24 17:06:36 So - testcloud? 17:06:43 roshi, ^^^^ 17:06:51 did we ever finish the discussion about cloud-init alternatives? 17:06:59 jzb, nope 17:07:00 testcloud package was approved, first updates were submitted yesterday 17:07:11 YAY :) 17:07:19 last i checked, it wasn't in updates-testing yet, though 17:07:29 jzb, also networkd (I will start a separate thread for that). 17:07:43 jzb, and motd for cloud images. 17:07:56 kushal: yes 17:07:59 kushal: thanks 17:08:26 I had an idea 17:08:30 is anybody interested in investigating the Flamingo and core-OS cloud-init in go as alternatives? 17:08:35 dustymabe: yeah? 17:08:41 jzb, we will have automated tickets filed for test failed images, can we have that as a feature ? :p 17:08:45 jzb: well I'll go with your item 17:09:07 I'm working with Tamer and we are talking with coreos guys about new cloud-init 17:09:18 dustymabe, cool 17:09:26 possible new organization for coreos-cloudinit + flamingo 17:09:38 will be a new baby that both coreos + us support 17:09:41 dustymabe: is that likely to fit into the F24 timeframe? 17:09:50 jzb: probably not 17:09:52 that'd be a spiffy feature to have 17:09:54 damn 17:09:55 :-( 17:09:55 would rather have that for f25 17:09:59 I think 17:10:02 dustymabe, I am okay with that. 17:10:14 jzb: I just don't want to commit to it before we have real traction 17:10:20 but we are working on it 17:10:30 dustymabe, also do you know what is the current status of flamingo? 17:10:42 dustymabe, then let us do it for f25 17:10:47 kushal: I think it works (maybe not feature complete) 17:10:53 okay 17:11:14 #action Kushal will write to list about #143 (f24 features) 17:11:18 but I think the result of the re-organization will probably mean the new codebase re-uses a lot of what coreos already has 17:11:28 We should also look into F25 ideas 17:11:37 kushal: sounds good 17:11:44 can you write that down on an F25 ideas page? 17:11:49 dustymabe, I will 17:11:57 we had sort of a dearth of new ideas for F23 17:12:04 I'm hoping we have more to show off in F24 17:12:06 dustymabe, put an action item on me :) 17:12:07 jzb: so I had one 17:12:12 jzb, :) 17:12:12 dustymabe: shoot 17:12:18 which is not exactly a new feature.. but is an interested idea 17:12:26 so vagrant has support for "remote" boxes 17:12:32 so like vagrant aws, etc 17:12:48 thought it might be interesting if we released some of those for fedora 17:12:59 dustymabe, Infra or the real app? 17:13:14 kushal: I'm not sure I understand the question 17:13:22 oh 17:13:43 dustymabe: what would releasing those look like (and/or require)? 17:13:46 basically you use vagrant-aws plugin to bring up instance and then vagrant ssh to it 17:13:47 dustymabe, is this something where I have the Vagrantfile in the local system but the instances will be on remote system? (or both remote) 17:13:54 ah 17:14:04 right 17:14:05 dustymabe: so we need to package the plugins, basically? 17:14:17 yup 17:14:18 jzb: yeah. I asked josef stribny about it 17:14:21 I think he is working on it 17:14:35 dustymabe, We have to ask him if he is okay this as a change :) 17:14:38 we would need to package the plugins and also release "boxes" that have the metadata 17:14:56 kushal: so I thought I would float it as an "idea" :) 17:15:04 to see what you guys think 17:15:10 dustymabe, say a *good* idea :) 17:15:38 dustymabe: +1 17:15:39 I think it's a good idea 17:15:53 ok. there is one other thing I am working on 17:16:13 which is a better way to have synced folders in vagrant (across providers and across OS) 17:16:17 which uses sshfs 17:16:32 I don't think this is a *change* for fedora 17:16:42 but just thought I would throw it out there 17:17:04 brb 17:17:09 I *think* this falls within the realm of the Cloud WG ... but maybe not -> https://fedoraproject.org/wiki/Changes/OpenShiftOrigin 17:17:30 maxamillion: nice 17:17:31 maxamillion: definitely 17:17:33 yes 17:17:38 maxamillion: what sort of help do you need there? 17:17:40 I have OpenShift v3 running on Fedora right now, but SELinux isn't happy (known policy issues, BZs are filed) 17:17:50 * tflink is going to be AFK for a bit, sorry about the timing 17:17:52 jzb: nothing at the moment, package is currently under review and I'm working through that 17:18:11 maxamillion: it's just one package? 17:18:17 jzb: yup 17:18:26 maxamillion, ah, golang++ 17:18:29 jzb: it's written in golang, everything is bundled and statically linked 17:18:31 maxamillion: does origin also give us clustering? 17:18:39 or is this single-host focused? 17:18:43 jzb: clustering 17:18:48 woot 17:19:16 Anything else? 17:19:22 jzb: I added support upstream for the ansible based installer to handle fedora also https://github.com/openshift/openshift-ansible 17:19:55 jzb: specifically with this install path https://github.com/openshift/openshift-ansible/blob/master/README_origin.md (I haven't tested others) 17:20:12 bah, I need to add a note about Fedora in there 17:20:18 ok.. any other items for changes? 17:20:19 https://github.com/openshift/openshift-ansible 17:20:22 https://github.com/openshift/openshift-ansible/blob/master/README_origin.md 17:20:23 before we move on 17:20:47 in the cloud-init discussion we also talked about "rich boot process" 17:20:47 dustymabe, nothing from me :) 17:21:02 I'm not sure how far along that is, though - and it takes over after cloud-init AIUI 17:21:17 jzb: more details about that and an open discussion would be nice 17:21:18 we can touch on that next week, I need to follow up 17:21:28 yup 17:21:32 ok moving on 17:21:44 #topic Vagrant-libvirt Failed for Fedora-Cloud-Atomic 23 17:21:50 #link https://fedorahosted.org/cloud/ticket/141 17:21:58 so I just sent a new patch to the list for this 17:22:10 my previous patch affected more than just the vagrant box 17:22:30 this one only affects the vagrant box, but we should refactor some of the other code for F24/rawhide 17:22:45 kushal: can you apply the patch and get a new build maybe? 17:22:48 that we can test 17:23:15 dustymabe, yes, I can 17:23:21 dustymabe, a local build? 17:23:24 sweet. if you respond in theticket I can test it out 17:23:39 kushal: local works fine.. if you can push it up and let me download to test 17:23:46 thanks 17:24:03 dustymabe, I will 17:24:21 kushal: there is also this https://fedorahosted.org/rel-eng/ticket/6304 17:24:31 please comment if you have an opinion 17:24:37 ok moving on 17:24:47 #topic Fedora Cloud FAD (late 2015/early 2016) 17:24:57 #link https://fedorahosted.org/cloud/ticket/115 17:25:09 so I think some people here will be at FOSDEM 17:25:18 dustymabe, devconf should be better :) 17:25:31 I'm not saying that we should make this FAD happen then, but I think we should at least get together to talk while we are close by 17:25:33 actually after devconf if possible 17:25:49 yeah I think we should do the FAD separately 17:26:21 dustymabe +1 17:26:22 dustymabe: I think there will be some folks meeting after 17:26:42 kushal: are you attending devconf? 17:26:47 jzb: ok let's make sure whoever happens to be there gets together 17:26:54 jzb, hopefully yes if budget permits 17:27:23 ok next ticket 17:27:29 #topic https://fedorahosted.org/cloud/ticket/115 17:27:35 #topic Migrate all Dockerfiles / Images to systemd where possible 17:27:43 #link https://fedorahosted.org/cloud/ticket/121 17:28:31 all of these dockerfiles tickets.. I might group them together at the end of something 17:28:37 does anyone have any comments on this one? 17:29:04 moving on.. 17:29:16 #topic make docker archived image get imported with lowercase tag 17:29:22 #link https://fedorahosted.org/cloud/ticket/131 17:29:33 imcleod: I think you submitted a patch for this 17:29:43 it's in imagefactory now? 17:30:48 ok well that is the status.. there was a PR 17:30:52 moving on 17:31:00 #topic Fedora 23 Retrospective 17:31:05 #link https://fedorahosted.org/cloud/ticket/135 17:31:14 jzb: kushal.. have anything for this? 17:31:34 jzb, so few of the volunteers are having exams, and two are trying out things regularly. 17:31:45 kushal: we noticed volunteers last week, thanks :-) 17:31:59 kushal: do we need to do any more calls for volunteers and/or maybe testing guidebooks? 17:32:10 i'm also getting up to speed on tunir tests 17:32:11 kushal: did you get an email from zeeshan. he is a new volunteer that is looking to contribute 17:32:13 kushal: we can always put stuff up on the community blog for Fedora now 17:32:28 jzb, yes, we will need few more posts, I will do those 17:32:34 jzb, put me on action items 17:32:44 dustymabe, yes, but I has on leave, I will reply tonight 17:32:49 or tomorrow early 17:32:53 kushal: you're a chair, you can add your own actions if you like ;-) 17:32:56 nzwulfin_afk, yay :) 17:33:06 jzb, still eating in one hand and typing in other 17:33:19 sorry dinner time clash 17:33:31 #action kushal to do more posts for attracting/guiding volunteers 17:33:36 done :) 17:33:40 dustymabe: you beat me to it. 17:33:41 dustymabe, thanks 17:33:50 ok done with this ticket? 17:33:53 the only other thing I would add to this is 17:34:08 we just need a better "story" for F23 than what we had for F22 17:34:14 er, sorry - f24 / f23 17:34:29 I don't know what that means 17:34:30 jzb: from a marketing perspective? 17:34:32 overall - from the cloud perspective, F23 wasn't a strong release. 17:34:48 maxamillion: it means we had very little to talk about in terms of new features or anything to show. 17:35:05 jzb: fair 17:35:09 maxamillion: we didn't have many changes filed for F23 and the ones we did have didn't make it. 17:35:27 jzb, true 17:35:28 but per the earlier discussion I think we're looking much better out of the gate so far. 17:35:37 definitely 17:35:41 especially if we have Origin packaged, etc. 17:35:53 #EOM 17:35:56 dustymabe: thanks 17:35:58 jzb: agree that is something that will be nice to talk about 17:36:08 jzb, and we also pushed down things for f24, this time we should target f25 from right now. 17:36:45 (also reminder - with two-week releases we can slip things in earlier for Atomic than F25, right?) 17:36:57 jzb, yes 17:37:00 jzb: define slip? 17:37:11 and you mean f24? 17:37:18 dustymabe, put out a new feature in open 17:37:22 hmm 17:37:31 I don't know if I like that 17:37:44 so for example.. slipping in networkd? 17:38:03 dustymabe: isn't that the advantage of doing a different release cycle? 17:38:07 well, one advantage. 17:38:24 well. I think the problem is that we aren't actually a rolling release at this point 17:38:33 dustymabe, For Atomic we are. 17:38:42 ]not really 17:38:52 I thought the point was to get people patches, not systemwide changes. 17:39:11 dustymabe: not really? 17:39:23 maxamillion: at least I don't think so 17:39:25 gholms, it makes sense to release the system wide thing later, but if required we can do that. 17:39:44 I think appropriate rolling release would mean we never "jump" 17:39:51 dustymabe: I was just curious by what you meant by that ... I think "rolling release" can be quantified/defined in different terms ... was just curious 17:39:54 I assume that the added pkgs would at least need to be in Fedora proper, so they'd be subject to some constraints 17:39:57 dustymabe: ah, gotchya 17:40:00 dustymabe: yeah, that's fair 17:40:10 dustymabe: we're kind of "rolling minor release" I guess 17:40:17 I'm +1 to never jumping :) 17:40:19 maxamillion: right 17:40:29 details, though :) 17:40:30 but that also means we can't just slip in a change to the network stack 17:40:40 or at least I wouldn't be comfortable doing that 17:40:42 dustymabe, for netwokd nope :) 17:41:00 or changing cloud-init to flamingo 17:41:02 dustymabe, I would prefer to put networkd as part formal f24 release 17:41:02 etc.. 17:41:06 right 17:41:19 so that is what I mean when I ask "what do you mean by slip in features" 17:41:23 :) 17:41:41 I'm not sure that we're really benefitting from a two-week cycle if we still hold back for major changes based on a six-month release cycle. 17:41:51 if something is ready mid-stream, we should go with it. 17:42:09 jzb: so what I think we should do 17:42:10 is 17:42:19 If we don't currently include, say cockpit-kubernetes, but that pkg is in fedora stable, putting that in the image would be an added feature 17:42:26 go with what we have now.. but in the future target a full rolling release 17:42:47 jbrooks: that is something I would be ok with I think 17:42:59 jzb, for something like networkd we should get the details for testing network stack from QA, and get our whole volunteer effort to test out before releasing. 17:43:00 dustymabe: when is the future? :-) 17:43:10 As QA is overloaded. 17:43:16 kushal: we definitely need to be testing, I agree 17:43:27 kushal: and we can't load it all on QA, definitely 17:43:43 jzb: not sure on when future is 17:43:48 jzb, that was one - point came up during F23 discussion. 17:43:50 I guess we would have to convince other people of it 17:43:52 so I think the real thing here is - what's our plan for having more robust QA from the cloud WG? 17:43:54 dustymabe, Future is now. 17:44:11 kushal: I thought now was the present. :-) 17:44:21 We could start a rolling side project, maybe a rawhide release for kicking the tires on the idea 17:44:37 Given enough interested hands to do the work 17:44:49 jzb, training and getting more volunteers for Cloud WG 17:44:56 jbrooks: yeah. I talked about that with maxamillion before I think 17:45:23 * jzb wonders how many is "enough"? 17:45:30 jbrooks, that sounds like a proper six month thing (the current model), but atomic we can test out things faster if we want to 17:46:01 jzb, :) 17:46:06 so we already have rawide atomic tree 17:46:38 maybe we could have rawhide -2 weeks 17:46:40 or something 17:46:48 called it cookedhide 17:47:00 Or build and test nightly, and only release if it passes 17:47:12 jzb, dustymabe for Vagrant image testing by volunteers, it would be great to access to a bare metal system where people can login and test 17:47:14 only packages that have been in rawhide two weeks without changing make it in :) 17:47:23 jzb, the volunteers are using Fedora cloud for qcow2 image testing 17:47:31 dustymabe: what would rawhide -2 weeks give us vs latest rawhide pending we have automated testing to verify at least some baseline of functionality? 17:47:39 kushal: tflink has been working on beaker for fedora I think 17:47:47 but can not test vagrant images except one person who is in Canada 17:48:09 tflink++ 17:48:09 kushal: Karma for tflink changed to 3 (for the f23 release cycle): https://badges.fedoraproject.org/tags/cookie/any 17:48:17 maxamillion: I guess that pending part is the concern 17:48:24 we don't really have them now do we? 17:49:19 anyway.. tl;dr we don't "really" have rolling release right now because we are based on f23 17:49:21 dustymabe: AutoCloud is testing cloud images, we're just not doing anything with them yet 17:49:44 maxamillion, it is testing the vagrant images too :p 17:49:54 Now may not be the time, but on the f23 and testing theme, I just confirmed a regression in our current atomic ISO image -- we're back to loopback storage 17:50:13 ah 17:50:18 wut 17:50:19 sigh 17:50:25 jbrooks, the iso image is getting tested in OpenQA iifrc 17:50:32 kushal: right, I just mean Cloud and Atomic are being tested 17:50:40 maxamillion, oh okay 17:50:42 ok 10 minutes left 17:50:49 We need to put in a test to ensure we're not using loopback lvm 17:50:58 moving to next topic 17:51:12 #topic Producing 2 week atomic images 17:51:19 #link https://fedorahosted.org/cloud/ticket/139 17:51:27 good job team for getting this out 17:51:35 after the next release I will close the ticket 17:51:45 what is the status for next release 17:51:51 what do we need people to do? 17:51:54 maxamillion: ^^ 17:52:34 I think jzb is scheduling a go/no-go meeting ... or someone was going to 17:53:01 dustymabe: really, people just need to contribute test cases where they are missing ... most everything else is done there 17:53:08 kushal: thoughts? ^^^ 17:53:23 maxamillion: a good way I have seen to do test cases is to look at the bug reports as they come in 17:53:29 and base test cases off of them 17:53:46 I've started opening issues for new test cases based on the bug reports 17:53:55 maxamillion: when would we need to do the go/no-go 17:54:05 jzb: please include me on the go/nogo and possibly send to cloud list as well 17:54:06 I think it was up in the air, but I can schedule & run it for the time being 17:54:09 I need someone to open the tickets, we can write down the tests. 17:54:22 dustymabe: I'll definitely send to cloud list :-) 17:54:25 kushal: is Issue against tunirtests good enough? 17:54:31 dustymabe, Yes, 17:54:33 jzb: thinks 17:54:36 sigh.. 17:54:38 thanks 17:54:46 It's one of those days 17:54:50 dustymabe, you started the trend :) 17:55:13 ok 17:55:15 next ticket 17:55:27 #topic Produce updated cloud base images monthly 17:55:29 dustymabe, I mean other than you we need more upstream/volunteer to file the tickets 17:55:35 #link https://fedorahosted.org/cloud/ticket/138 17:56:12 so should we target this for the next two week atomic release? 17:56:20 or is this something we want to put off for now 17:56:30 dustymabe, we should release a base image 17:56:52 dustymabe, we are getting them tested. 17:57:00 dustymabe: do we know what state we're in generally here? 17:57:15 I'm fine with doing a twofer and having a go/no-go for both 17:57:16 jzb: I think it is the exact same state as the 2 week atomic stuff 17:57:29 but on a 4 week cadence 17:57:41 dustymabe: might as well take advantage of folks' attention and do it all at once 17:57:49 kushal: what's the latest you're available for meetings? 17:58:00 in UTC or Eastern 17:58:13 jzb when do you want it? 17:58:37 jzb, say upto 6pm UTC to start the meeting 17:58:40 maxamillion: we'd need to do the go / no-go Friday or Monday? 17:58:49 jzb: I don't know, ask mmcgrath 17:58:53 jzb, please skip this Friday :p 17:58:58 jzb, Monday please 17:59:11 kushal: will see what I can do 17:59:12 I want my friday evening without meetings :) 17:59:26 kushal: we all have dreams ;-) 17:59:54 jzb, 6pm means already 11:30pm here 18:00:00 j:D 18:00:02 :D 18:00:17 It is time 18:00:20 dustymabe, ^^ 18:00:31 ok guys 18:00:39 #topic openfloor 18:00:47 if you have anything we better make it quick :( 18:01:10 otherwise I am setting fuse 18:01:13 10 18:01:14 Let us discuss on list 18:01:18 dustymabe, 3 18:01:18 :p 18:01:20 9 18:01:22 8 18:01:24 7 18:01:26 6 18:01:28 5 18:01:29 2 18:01:29 4.5 18:01:30 4 18:01:32 3 18:01:35 0.9 18:01:38 :) 18:01:40 #endmeeting