17:01:32 #startmeeting fedora_atomic_wg 17:01:32 Meeting started Wed Sep 13 17:01:32 2017 UTC. The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:32 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:01:32 The meeting name has been set to 'fedora_atomic_wg' 17:01:35 #topic roll call 17:01:39 .hello dustymabe 17:01:40 .hellomynameis kushal 17:01:40 dustymabe: dustymabe 'Dusty Mabe' 17:01:41 .hello sayanchowdhury 17:01:43 kushal: kushal 'Kushal Das' 17:01:46 sayan: sayanchowdhury 'Sayan Chowdhury' 17:01:48 .hello miabbott 17:01:52 miabbott: miabbott 'Micah Abbott' 17:01:53 .hello sinnykumari 17:01:54 ksinny: sinnykumari 'Sinny Kumari' 17:02:23 roshi: around today for meeting? 17:02:27 .hello jasonbrooks 17:02:28 jbrooks: jasonbrooks 'Jason Brooks' 17:02:36 .hello roshi 17:02:37 roshi: roshi 'Mike Ruckman' 17:02:48 how's it going folks? 17:03:11 roshi: good good, glad to see you :) 17:03:27 :D 17:03:32 * roshi is still around 17:03:36 just not tracking nearly as much 17:05:22 ok do we have everyone? 17:05:31 walters: jlebon 17:05:39 .hello maxamillion 17:05:39 maxamillion: maxamillion 'Adam Miller' 17:05:50 .hellomynameis walters 17:05:51 walters: walters 'Colin Walters' 17:06:05 #chair kushal sayan miabbott ksinny jbrooks roshi maxamillion dustymabe walters 17:06:05 Current chairs: dustymabe jbrooks ksinny kushal maxamillion miabbott roshi sayan walters 17:06:41 * dustymabe was waiting on tomas and eslobodo 17:06:43 .hello jlebon 17:06:44 jlebon: jlebon 'None' 17:06:49 #chair jlebon 17:06:49 Current chairs: dustymabe jbrooks jlebon ksinny kushal maxamillion miabbott roshi sayan walters 17:06:58 .hello eslobodo 17:06:59 eslobodo_: eslobodo 'Eliska Slobodova' 17:07:15 yay 17:07:19 ok let's get started 17:07:20 dustymabe: sorry for being late 17:07:25 eslobodo_: no worries 17:07:31 #topic previous meeting action items 17:07:59 jbrooks to submit asciibinder pkg for review 17:08:01 jberkus to follow up on partial container images from Flock 17:08:03 jbrooks to create issue for adding examples to container guidelines 17:08:18 where's jberkus when you need him? 17:08:21 :) 17:08:22 dustymabe, oh dang, I didn't do either of those -- the second I totally forgot 17:08:32 jbrooks: no worries 17:08:34 #action jbrooks to submit asciibinder pkg for review 17:08:41 #action jbrooks to create issue for adding examples to container guidelines 17:09:03 since josh isn't here... 17:09:09 #action jberkus to follow up on partial container images from Flock 17:09:19 ok moving on 17:09:23 He's at osssummit 17:09:36 #topic [Test Day] Proposed Dates for F27 Atomic Test Day 17:09:43 #link https://pagure.io/atomic-wg/issue/322 17:10:08 we had sumantrom ask us about a test day for FAH for 27 17:10:17 let's try to find a time where we can get together and test things 17:10:52 any ideas for a good time for this? 17:10:54 * roshi will try to do some testing at least 17:11:03 o/ 17:11:08 ttomecek: \o 17:11:11 #chair ttomecek 17:11:11 Current chairs: dustymabe jbrooks jlebon ksinny kushal maxamillion miabbott roshi sayan ttomecek walters 17:11:26 roshi: will you help us navigate wiki land? 17:11:45 gimme a date and I'll try to pull the things together 17:11:51 ttomecek: \o 17:12:07 proposals for dates anyone? 17:12:11 but still check in with the QA team, file a ticket and it'll be tracked at least 17:12:12 we aren't technically tied to beta 17:12:35 so we don't have to wait for beta to do testing (and i might even prefer that so we can get fixes in sooner) 17:12:38 * ksinny is available for test day till next week. will be away from work between 23rd sep to 2nd oct. 17:12:44 * dustymabe looks at schedule 17:13:06 Same with me (about the schedule ksinny shared) 17:13:11 2017-09-05 Beta Freeze (*) 17:13:12 2017-09-19 Beta Release (Target date) 17:13:14 2017-09-26 Beta Release (Rain date) 17:13:16 2017-10-10 Final Freeze (*) 17:13:18 2017-10-24 Fedora 27 Final Release (GA) (Target date) 17:13:20 2017-10-31 Fedora 27 Final Release (GA) (Rain date) 17:14:16 ehh - so let's just target next week then 17:14:19 anyone opposed? 17:14:26 not I 17:14:58 friday is probably best since people are usually busier with meetings during the other days of the week 17:15:14 ehh, or thursday 17:15:20 votes? 17:15:27 oh damn 17:15:40 just realized i'm going on a trip next thurs/friday 17:15:45 heh 17:15:48 *facepalm* 17:15:51 heh 17:15:58 I'm fine with whatever - though there's potential for window in there where I won't be around 17:16:04 just as an FYI 17:16:07 roshi: of course 17:16:10 how about next wed 17:16:17 wednesday then? 17:16:23 yeah, just whenever 17:16:36 wed works 17:16:41 anyone opposed? 17:16:47 works for me 17:16:56 works for me as well 17:17:22 ok 17:17:42 #agreed, will try to shoot for a test day for next friday for Fedora Atomic Host 17:18:08 dustymabe: wednesday, I believe 17:18:11 dustymabe, next fri? 17:18:19 #undo 17:18:19 Removing item from minutes: AGREED by dustymabe at 17:17:42 : , will try to shoot for a test day for next friday for Fedora Atomic Host 17:18:24 :) 17:18:30 agreed, will try to shoot for a test day for next wednesday for Fedora Atomic Host 17:18:34 #agreed, will try to shoot for a test day for next wednesday for Fedora Atomic Host 17:18:38 :) 17:18:39 next wednesday is a short timeline to get the word out 17:18:42 real time collaboration 17:19:12 roshi: yeah 17:19:28 so i'm out th/fr and ksinny and kushal are out the following week 17:19:40 I am also out the following week 17:19:49 sayan: :( 17:19:53 holiday in India? 17:20:08 dustymabe: yup, long festival 17:20:26 going hometown 17:20:32 +1 17:20:35 roshi: thoughts? 17:20:52 obviously we don't need everyone present for a test day, but let's try to find a good balance 17:20:58 ksinny: very cool, have fun! 17:21:09 maxamillion: thanks :) 17:21:11 I'll be around in passing, but not full time to help people out 17:21:15 $dayjob and all 17:21:22 roshi: of course 17:21:53 hey, i forget how to inject topics, but let's add https://pagure.io/fedora-kickstarts/pull-request/266 17:22:18 walters: ok 17:22:56 roshi.. advice on best path forward? 17:23:15 should be able to basically reuse the last atomic test day 17:23:30 unless there's some new thing you want tested 17:23:47 someone needs to write up something for the community blog 17:23:59 and twit about it on the tweeter and such :) 17:24:00 roshi: mostly from last time. and maybe we can get something from jbrooks/strigazi for testing kube in system containers 17:24:11 roshi, mostly asking for advice on date 17:24:23 i.e. do you think next wednesday is too soon? 17:24:39 if you want to add tests, like for kube, I'd say give yourselves 2 weeks to get everything in order 17:25:03 I'd say it's too soon, but it all depends on how hard you want to push 17:25:25 ok i'll add all of this info to the ticket 17:25:51 * dustymabe wishes we didn't have too choose between "too soon", "many people being AFK", and "pending release" 17:25:52 :) 17:25:58 s/too/to 17:26:08 nature of the beast my friend :) 17:26:11 right 17:26:41 ksinny: sayan: kushal: if we do it the week you are away we'll make sure to have some material in place for you to run through before/after your vacay 17:26:55 dustymabe, Okay. 17:26:59 next topic 17:27:03 dustymabe: sure 17:27:05 for kube tests, I'd do: http://www.projectatomic.io/blog/2017/09/running-kubernetes-on-fedora-atomic-26/ 17:27:09 but w/ f27 17:27:17 ttomecek: eslobodo do you mind me bringing up your work as a topic? 17:27:51 dustymabe, no problem 17:28:03 #topic work from ttomecek eslobodo on fedora containers 17:28:11 Okay 17:28:18 ttomecek: and eslobodo have been doing some work on getting our container offerings in better shape 17:28:31 ttomecek: eslobodo can you speak about it and let people know what you've been up to? 17:28:59 one of the goals of our team (userspace containerization) is to containerize more things 17:29:20 and improve existing image: hence we did tools, systemtap and vsftpd recently 17:29:40 Eliska is working hard on improving container best practices 17:30:03 other goal is to support modularity, so we're also turning some of the images to use modules 17:30:18 pretty much all of our work is here: https://github.com/container-images 17:30:43 eslobodo: about modularity.. how is that going to work with our current container image release process? 17:30:46 we are also trying to come up with an easy way to test containers and images: I hope we'll be able to share something soon 17:30:57 where we have images under a namespace like f25/httpd ? 17:31:25 dustymabe, for boltron, we created a new modular base image which lives in its own namespace 17:31:40 I'm not sure if this is gonna be the same for modular f27 17:31:46 the base image has it's own tag, not it's own namespace 17:31:48 right? 17:31:59 #info ttomecek and eslobodo's team (userspace containerization) is trying to containerize more things and improve existing images. The did tools container, systemtap and vsftpd recently. Check them out! 17:32:13 maxamillion, yup: registry.fedoraproject.org/f26-modular/boltron 17:32:18 ah ok 17:32:19 awesome 17:32:28 +1 17:32:33 I think the one on docker hub is just a tag 17:32:44 I can't ever remember all this stuff, so many moving parts 17:32:52 ttomecek eslobodo thanks 17:33:25 dustymabe, you're welcome (back to dinner :D) 17:33:26 ttomecek: eslobodo. if you don't mind please put any big initiatives you are working on into the atomic-wg issue tracker. this will help everyone know what is being worked on and also if they can help 17:34:27 dustymabe, sounds like a good idea, I'm just worried about keeping it up to date 17:34:53 ttomecek: :) "big initiatives" 17:35:07 hopefully not too many of them 17:35:26 we're also trying hard to keep everything in as few places as possible :) but we'll try :) 17:35:28 "We're working on FOo and here's a TL;DR:" is probably enough :) 17:35:38 thanks again ttomecek and eslobodo 17:36:02 #topic update partitioning model in Fedora Atomic Host 17:36:11 #link https://pagure.io/fedora-kickstarts/pull-request/266 17:36:19 walters: ^^ 17:36:43 i'm going to test that PR again right now 17:37:12 walters: so this PR makes it so that the cloud images come with their entire disk utilized ? 17:37:30 there was a ton of discussion on this but i think we should (after I test) merge it and we can adjust on feedback 17:37:35 dustymabe, yeah 17:37:53 +1 17:38:02 walters: I'm +1 for rawhide 17:38:15 you're really thinking we wouldn't do this for 27? 17:38:39 walters: i'm just thinking that we should test it first in rawhide 17:39:06 okay 17:39:07 i don't want to merge/revert in f27 right now because of 'freeze' 17:39:33 the problem with rawhide as dev stream is without a sane ability to revert it constantly breaks 17:40:00 it's always been that way of course 17:40:06 walters: yeah. good news is that it seems to be working right now :) 17:40:13 https://github.com/projectatomic/atomic-host-tests 17:40:16 dustymabe: today ;) 17:40:58 walters: i still think we should consider scrapping LVM at some point 17:41:03 but a-h-t don't gate RPMs 17:41:19 actually they don't affect the FAH release streams directly in any automated way at all 17:41:20 right, but i'm just pointing that it's passing tests 17:41:37 anyways yes 17:41:39 so rawhide seems in an ok state for right now, so we should be able to test this 17:41:52 fwiw, a-h-t doesn't have any tests that would find an issue with the new partitioning scheme 17:42:00 other than 'oops, it doesn't boot' 17:42:05 miabbott: correct 17:42:17 ok next topic 17:42:25 #topic packages in fedora 27 atomic host 17:43:05 so the diff between f26 and f28 includes removing "kubernetes", "etcd","flannel""glusterfs", "glusterfs-fuse", ceph-common 17:43:22 we are +1 to keeping kubernetes out (thanks jbrooks/strigazi/yzhang) 17:43:41 i think the glusterfs stuff at least needs to go back in (glusterfs/kubernetes work) 17:43:41 +1 17:43:48 does it? 17:43:52 strigazi wanted to also keep ceph-common 17:44:06 maxamillion: what's the question? 17:44:08 I thought you could run glusterfs containerized (I'm pretty sure that's how openshift does it) 17:44:24 can't we just layer in those pkgs afterwards? seems like most users don't have a use case for them 17:44:28 maxamillion: right, but I think it uses glusterfs-fuse on the host for some stuff 17:44:38 dustymabe: hrmmm.... 17:44:48 maxamillion: i could be wrong 17:44:56 dustymabe: also what miabbott said, can't we just layer that stuff in based on the user/admin's preferences/desires? 17:44:57 i will open a ticket and ask the maintainer to comment 17:45:05 keep the base lean, let people use it however they want 17:45:06 +1 to layering 17:45:30 +1 lean base 17:45:57 so the only problem i have with that at this point is we haven't taught any tools to do package layering 17:46:07 like openshift-ansible 17:46:26 Does openshift-ansible need any of those pkgs? 17:46:41 jbrooks: i'll open a ticket and get the maintainer to comment on that 17:46:53 i think it does, but I don't want to be wrong 17:47:19 jbrooks: and of course it only matters if you try to use gluster as part of openshift-ansible 17:47:26 make sense? 17:48:01 What I find weird is that we've already decided to remove these -- they were removed, it was committed, and now we have to keep talking about adding them back 17:48:02 #action dustymabe to open ticket regarding gluster/ceph packages in FAH 17:48:13 dustymabe: we can add that though 17:48:33 Layering is fine, it'd be good to have an ansible module that knows about it 17:48:46 Same w/ system containers 17:48:58 any volunteers for a package layering ansible module? 17:49:18 the openshift-ansible team is quick to merge things 17:49:39 dustymabe: I'll look into it 17:49:46 maxamillion: yes they are, but what changes to openshift ansible would we need to make? 17:49:56 dustymabe: don't know just yet 17:49:58 https://github.com/openshift/openshift-ansible/blob/master/roles/etcd/tasks/system_container.yml 17:49:59 i thought we'd talked about teaching the generic ansible `pkg` module how to do this 17:50:07 they just use command: 17:50:09 maxamillion: also, that 'module' could use some design discussion :) 17:50:21 https://github.com/ansible/ansible/issues/21185 17:50:28 for the system containers, we could do the same w/ layering 17:50:38 dustymabe: we can setup an open invitation bluejeans call to hash it out 17:50:48 sure 17:51:02 ok so that is for the gluster/ceph stuff 17:51:08 flannel/etcd ? 17:51:20 We have system containers for those 17:51:24 i'm +1 for keeping flannel out 17:51:27 +1 17:51:39 i could see etcd as a general purpose "cluster" thing that could be useful left in 17:51:53 but either way works for me 17:51:54 But just as easily run in a container 17:51:58 or with layering 17:52:13 containers or gtfo 17:52:22 haha 17:52:25 heh 17:52:35 i kid, i kid 17:52:41 It's out now, the Q is do we need it back in, the answer is no 17:52:46 but i'm +1 to removing etcd 17:52:53 jbrooks: 'now' is relative 17:53:09 last 2WA release has it 'in' 17:53:17 I refer to the actual current state of the f27 json 17:53:18 +1 to removing etcd also 17:53:19 right? 17:53:25 remove everything not absolutely necessary 17:53:28 jbrooks: indeed, you are right 17:53:36 maxamillion: tmux ? 17:53:38 :) 17:53:55 now that is an interesting topic =) 17:54:13 let's not be hasty 17:54:34 dustymabe: leave it out 17:54:46 I can layer it in for my needs, but not everyone wants it 17:54:59 #info decided to keep kube/etcd/flannel out of the ostree in f27. will visit topic of ceph/gluster in an issue where we can ask the maintainer if openshift-ansible needs it to be installed on the host. If we teach ansible how to package layer then removing more things could happen. 17:55:22 fair? 17:55:27 do we still need to reboot after a package layer operation? 17:55:31 dustymabe: yes, fair +1 17:55:37 maxamillion: short answer yes 17:55:41 i guess that means we have to drop 'screen' too? 17:55:46 dustymabe, we should point out that ansible doesn't know how to do system containers, either 17:55:47 long answer, no, but it's experimental 17:55:50 it just runs the command 17:56:10 So, it knows just as much how to layer 17:56:28 miabbott: yeah, I'd throw out screen too 17:56:35 savage 17:56:45 strace? tcpdump? 17:56:56 I *live* in tmux, but still think it should be out 17:56:57 ok anyone opposed to moving to next topic? 17:57:02 yup, strace and tcpdump out 17:57:19 actually ... no, maybe 17:57:31 #topic open floor 17:57:32 debugging if something goes sideways and you can't install packages that might be useful 17:57:49 anyone with anything for open floor ? 17:57:53 not I 17:58:02 any modularity news/questions? 17:59:03 i think that will be a background topic...hopefully pick it up more again next week 17:59:39 cool cool. I'll close this out in 60 seconds 18:00:27 #endmeeting