17:03:51 #startmeeting cloud_wg 17:03:51 Meeting started Wed Nov 4 17:03:51 2015 UTC. The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:03:51 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:03:59 .fas rtnpro 17:03:59 rtnpro: rtnpro 'Ratnadeep Debnath' 17:04:01 I am not here (mostly) :( 17:04:04 #topic rollcall 17:04:07 kushal: :-( 17:04:12 .fas jasonbrooks 17:04:12 jbrooks: jasonbrooks 'Jason Brooks' 17:04:14 .fas rtnpro 17:04:16 rtnpro: rtnpro 'Ratnadeep Debnath' 17:04:19 .hello scollier 17:04:22 scollier: scollier 'Scott Collier' 17:04:51 #chair scollier rtnpro jzb jbrooks 17:04:51 Current chairs: dustymabe jbrooks jzb rtnpro scollier 17:04:53 .fas jzb 17:04:53 jzb: jzb 'Joe Brockmeier' - jzbl 'Denise Gant' 17:04:56 sigh 17:05:02 .hellomynameis jzb 17:05:03 jzb: jzb 'Joe Brockmeier' 17:05:13 haha jzbl?? 17:05:14 .hello maxamillion 17:05:14 maxamillion: maxamillion 'Adam Miller' 17:05:20 #chair maxamillion 17:05:20 Current chairs: dustymabe jbrooks jzb maxamillion rtnpro scollier 17:05:27 * gholms waves 17:05:56 #chair gholms 17:05:56 Current chairs: dustymabe gholms jbrooks jzb maxamillion rtnpro scollier 17:06:09 everyone gets a chair :) 17:06:18 * gholms shrugs, takes a seat, then :) 17:06:27 .fas sayanchowdhury 17:06:28 Guest38814: sayanchowdhury 'Sayan Chowdhury' 17:06:46 sayan do you want to fix that name? 17:07:14 # last week action items 17:07:20 I am going to do them by person this time 17:07:25 dustymabe, having network issues, too much lag :( 17:07:32 dustymabe to integrate cloud WG and 2 week atomic efforts 17:07:33 dustymabe to close ticket 94 and open a couple more specific tickets 17:08:16 I failed on both of these because of a lot of other stuff going around.. will carry them forward 17:08:27 #action dustymabe to integrate cloud WG and 2 week atomic efforts 17:08:33 #action dustymabe to close ticket 94 and open a couple more specific tickets 17:08:40 ok other action items: 17:08:46 jzb to write up a magazine article on bootstrapping if the pitch is accepted for the magazine 17:08:48 after F23 GA, jzb to schedule a meeting between the groups for a strategy discussion 17:08:50 jzb to start discuss thread on cloud/atomic/server wg stuff 17:09:03 I did that! 17:09:07 jzb: I wrote an article on ansible so I think that one is taken care of? 17:10:13 dustymabe: Yes 17:10:40 ok and the other items were taken care of by your mail list thread? 17:10:51 dustymabe: I think so, yes 17:10:53 dustymabe: I need to follow up with you about the integration with two week atomic at some point, I'd like to know what the scope of that is ... there's a lot going on around all of that at the moment 17:10:59 I don't think we've actually come to any conclusion 17:11:21 maxamillion: we'd love to have everyone involved join us here 17:11:48 we should probably be all in the same group 17:12:04 jzb: right 17:12:34 dustymabe: it's ReleaseEngineering, we cover all of Fedora, I don't think it's realistic for us to attend every single Fedora meeting ... we'd never get anything done 17:12:37 I really think it would be best to get interested parties in a room (and on video chat for those who couldn't make it) and hash it out 17:12:42 get everyone on the same page 17:13:20 my previous two statements were meant for jzb 17:13:29 I also don't think it's realistic to go "oh hey, we want to integrate with what you're doing ... come join us" ... should be the other way around 17:13:37 alright 17:13:52 maxamillion: ok. 17:14:17 let's talk about this soon.. whether they come to us or we go to them I think we should all be together 17:14:24 I'm interested in understanding fedora's atomic processes -- tell me what to read / where to show up and I'm there 17:14:35 at least for a meeting or something during the week 17:14:39 (Just a sec) 17:14:59 dustymabe: it's not a "us" or "them" ... I am part of "them" ... it's just a matter of organizational concern 17:15:20 dustymabe: we'll chat off-meeting, we'll get it sorted 17:15:51 I get it.. but if we put all of the burden on releng for delivering two week atomic then that kinda sucks right? and we are just sitting here making decisions that they have no part in (obviously you are part of "they" too) 17:16:22 I'm just saying let's integrate somehow.. ok let's talk after meeting 17:16:39 final action items are: roshi to put in for the magazine pitch 17:16:48 roshi: I'm not sure what that one was about 17:18:15 I actually think that was about ansible too 17:18:22 I looked at the logs 17:18:29 so I guess that one is taken care of 17:18:30 (Yes, dustymabe agree) 17:18:41 so I'm the only one with action items to carry forward :) 17:18:45 \o/ 17:18:54 (so far) 17:18:55 * dustymabe needs to learn to do his own AI and not others 17:18:58 dustymabe: wait, what about ansible? 17:19:07 maxamillion: fed mag post 17:19:13 dustymabe: oh ok 17:19:20 educating people on how to work around ansible+python3 issue 17:19:38 ok tickets 17:19:56 #topic Migrate all Dockerfiles / Images to systemd where possible 17:20:05 #link https://fedorahosted.org/cloud/ticket/121 17:20:47 scollier: praveenkumar ^^ 17:21:19 dustymabe: it should largely be a non-issue with Atomic 17:21:22 * scollier looks 17:21:54 dustymabe: so I made all changes and I think we can migrate other docker file also. 17:22:25 dustymabe: I will wait for scollier signal. 17:22:27 dustymabe, i have not seen a PR come in on that 17:22:37 praveenkumar, ^ 17:23:41 scollier: I will send PR soon, on ticket I just put my finding to open a discussion. 17:23:47 cool 17:23:49 thanks guys 17:23:51 praveenkumar, ack. thx 17:24:12 #topic fedora-dockerfiles: Clean up READMEs. 17:24:19 #link https://fedorahosted.org/cloud/ticket/122 17:24:32 coolsvap: ^^ 17:25:36 looks like he isn't here 17:25:36 dustymabe: I don't know if coolsvap is there, I will ask him to update on that ticket. 17:25:43 praveenkumar: thanks 17:25:46 .hellomynameis imcleod 17:25:56 welcome imcleod 17:25:59 #chair imcleod 17:25:59 Current chairs: dustymabe gholms imcleod jbrooks jzb maxamillion rtnpro scollier 17:26:10 imcleod: imcleod 'Ian McLeod' 17:26:17 #topic Document process for using Fedora-Dockerfile branches 17:26:23 #link https://fedorahosted.org/cloud/ticket/123 17:27:38 looks like adimania is not here this week 17:27:59 moving on 17:28:12 #topic Fedora-Dockerfiles examples for Kubernetes 17:28:18 #link https://fedorahosted.org/cloud/ticket/125 17:28:37 scollier: ^^ 17:29:02 dustymabe, was it ever settled where to put these? i thought there was a thread on it discussin. 17:29:26 i really don't knnow 17:29:43 dustymabe, ok, let me follow up on that then. 17:30:00 this sounds really close to just merging these in and making them either atomicapps or using something like helm from deis 17:30:11 same thing really 17:30:20 ok moving on 17:30:35 #topic https://fedorahosted.org/cloud/ticket/125 17:30:38 crap 17:30:42 #topic Working with the Server WG 17:30:49 #link https://fedorahosted.org/cloud/ticket/127 17:30:55 jzb sent a mail on this 17:31:00 progress pending?? 17:31:13 yeah, still ongoing conversation it seems 17:31:24 the thread is at least not stale so I'd call that progress 17:31:26 yeah, still ongoing 17:31:33 maybe we can co-locate a server+cloud FAD or something 17:31:39 ok moving on 17:31:48 #topic make docker archived image get imported with lowercase tag 17:31:55 #link https://fedorahosted.org/cloud/ticket/131 17:32:18 so we probably need to fix the tools 17:32:33 as well as update the docker html page for f23 to let them know? 17:32:53 https://getfedora.org/en/cloud/download/docker.html 17:33:19 huh? 17:33:32 maxamillion: lost? 17:33:35 dustymabe: IIRC, we found that this inherits from the "pseudo package name" that is used to build the image in koji. 17:33:50 dustymabe: what needs fixing about the website? 17:34:06 maxamillion: we should probably have the user tag the docker image with something they can use 17:34:14 dustymabe: oh ok 17:34:36 dustymabe: You showed a fix for the releng scripts but it was rather late in the cycle for F23 and may have had side effects we didn't know about. (dgilmore may comment) 17:34:53 yeah that's right. so I say we fix it for F24? 17:34:57 dustymabe: we also push to the docker hub, currently fedora:latest and fedora:23 are up there ... that was live about 45 minutes before Fedora 23 was officially released on the mirrors 17:35:14 * roshi just got back from an appt 17:35:15 dustymabe: I could fix this in Factory, and probably should. If uppercase tags are not acceptable then I should not be generating base images that contain them. 17:35:17 sorry for being late 17:35:26 roshi: *jeez* ;) 17:35:26 imcleod: right 17:35:30 can we open a bug for that? 17:35:43 dustymabe: I'll do it in github right now. 17:35:57 maxamillion: the ones in the hub are fine 17:36:19 imcleod: thanks.. then maybe put a link to the work in the ticket? 17:36:44 dustymabe: https://github.com/redhat-imaging/imagefactory/issues/359 17:36:58 dustymabe: This is also a good reminder that there's a backlog of issues from 2014 that I should close out. Sigh. 17:37:22 imcleod: :( 17:37:37 does someone want to take an AI to get the website updated? 17:38:57 dustymabe: usually I would, but PTO 17:39:15 #action dustymabe to get website updated for docker image tags issue 17:39:21 ok moving to open floor 17:39:26 #topic open floor 17:39:44 so vagrant images are in atlas.. and I'm going to do a fed mag post about it soon 17:39:48 dustymabe++ 17:39:55 we also need to do some work to hand this off to releng 17:40:16 https://fedorahosted.org/cloud/ticket/133 17:40:25 I opened that ticket to track it 17:40:59 I've been working on making a VM image that is self-replaceable... I have figured it out CentOS and Fedora so far (cloud-init part not yet) 17:41:25 spstarr self-replacable? 17:41:45 its not too complex, but I hope to turn this shell code and port to livemedia-creator 17:41:47 mattdm: yes 17:42:12 what does that mean? :) 17:42:43 dustymabe: that might be a problem 17:42:50 the VM image contains one partition, /boot which uses extlinux bootloader as a 'stub' a vmlinuz + initrd.img which inside contains the live fileysstem (install.img or squashfs.img) 17:42:56 dustymabe: we got ping'd by legal about Fedora images being in Atlas 17:43:05 dustymabe: need to follow up on that 17:43:19 maxamillion: I don't know why. I got a thumbs up from legal like 2 weeks ago 17:43:24 mattdm was on the mail 17:43:31 mattdm: so, you can upload this 'stub' VM and turn it into an AMI, when you reuse it again it installs the OS from a kickstart provided via cloud-init to anaconda 17:43:37 * mattdm confirms ack by legal 17:44:02 dustymabe: mattdm: dgilmore and I ended up in a different email thread just this past Friday about it ... I'll follow up off-meeting with everyone 17:44:10 maxamillion: ok 17:44:14 mattdm: so when the VM finishes install it contains the new OS and the original content is wiped out 17:44:20 spstarr: ah, nice. (we did a similar thing for provisiong systems at my last job, but via pxeboot) 17:44:48 here is some code: http://paste.fedoraproject.org/286928/46659078 17:45:00 mattdm: I think his use case was being able to somehow make this portable across clouds? 17:45:08 spstarr: ^^ was that right? 17:45:15 yes 17:45:35 cloud-init should be able to support X cloud provider and I am looking at a fallback to private cloud also 17:45:39 dustymabe: nevermind, it's good 17:45:46 so some systemd service addon or dracut plugin 17:45:48 dustymabe: you're right, legal signed off ... I'm a derp 17:46:02 this part im unsure of just yet, but working on that today 17:46:16 okay, whew. :) 17:46:16 maxamillion: no worries, i think there are mutliple channels going on 17:46:25 Fedora diverged quite a bit in the creation of the initrd image 17:46:39 not difficult though 17:46:52 spstarr: that's pretty cool. 17:47:11 (was this the next topic?) 17:47:16 (or did we get derailed?) 17:47:20 jzb: this is open floor 17:47:25 since we left an undocumented dracut option from history (Red Hat Linux) this makes this method possible 17:47:38 ah, I missed the "open floor" thing 17:47:39 dustymabe: yeah, I just went back and re-checked the email thread I was on and I was crossing mental streams ... apologies 17:47:41 * jzb has something 17:48:27 spstarr: that is interesting.. so is the goal to make an AMI that one then uses everytime they run on a cloud provider? 17:48:35 once i iron out these bits i'll look at porting to Fedora's tool to add options 17:48:35 * mattdm too but maybe the same thing or at least an aspect of it :) 17:48:36 or is the goal to run this every time you start an instance? 17:49:00 dustymabe: yes any provider, depends on what you want to do: you can use his VM image stub as a gold image creator within a cloud environment 17:49:19 spstarr: how big does the stub compress down to? 17:50:00 spstarr: no, what I am saying is: is the output of this an AMI/IMAGE that is then used to start instances, or is the goal to use this to modify a running instance into what you want? 17:50:12 and then use that instance? 17:50:19 so it works like this: run livemedia-creator -> specify size of disk image, it will get the boot.iso, extract the bits, patch initramfs with cloud-init / turn on systemd services --> patch kernel/initrd into stuf VM --> upload to cloud provider. 17:50:31 dustymabe: either or 17:50:43 it can be used as a bootstrap either case 17:51:17 so a bootstrap AMI --> when you create a new instance of the AMI --> that instance becomes a real VM -> make a new AMI --> use that AMI as 'gold image' --> production 17:51:54 so a use case is this: 17:53:30 1) I want to have a gold image based on revision of a kickstart.txt (passed in via cloud-init dynamically content or URL) from this stub image, I can build a Fedora 23 gold image.. when Fedora 24 comes out.. I can use that stub to make Fedora 24, 25, 26... etc 17:53:34 jzb: you might want to throw your item in and let that discussion go on too 17:53:44 or CentOS 7.1, 7.2, 7.3, 7.4.. 17:53:54 dustymabe: OK 17:54:02 https://fedorahosted.org/cloud/ticket/135 17:54:11 We need to have a Fedora 23 retrospective 17:54:21 (and then kick off F24 planning) 17:54:21 dustymabe: you basically are creating a template installer VM 17:54:42 I think it's safe to say that we could have done better for Fedora 23 17:54:49 spstarr: yeah.. but it's hard to promote that as an option for Fedora if we don't actually use it to create our release media 17:55:19 spstarr: you should ping imcleod in #fedora-cloud sometime and see what he thinks 17:55:20 I hate to lob something over the wall and run 17:55:24 (but I will) 17:55:30 he has a lot of history around installs 17:55:35 can we set a deadline to go through this in two weeks/ 17:55:36 ? 17:55:45 i guess the issue is Fedora has no 23.1, 23.2.. vs CentOS which has 7.x releases 17:55:49 * mattdm looks at calendar 17:56:03 CentOS is always CentOS-latest 17:56:05 and for Fedora anaconda changes per release 17:56:17 jzb: seems reasonable 17:56:23 so your stub VM that was using Fedora 22 to install Fedora 24 might not work... 17:56:34 spstarr: yeah, very likely won't 17:56:47 so this might be better used for CentOS 17:56:54 wait, why are we making stub images for AMIs? fedimg makes and uploads them constantly 17:57:05 http://fedimg.readthedocs.org/en/latest/services/ec2/ 17:57:06 maxamillion: we aren't as fedora 17:57:13 dustymabe: yes we are 17:57:17 dustymabe: every single compose 17:57:18 maxamillion: its invert... uploading 8GB of data each image costs $ 17:57:18 I'm gonna be at LISA next week and I'd like to participate in retrospective, so week _after_ that would be nice 17:57:19 this was a project spstarr is working on for personal 17:57:24 use 17:57:29 ohhhh ok 17:57:31 gotchya 17:57:35 spstarr: makes sense 17:57:44 this is uploading a stub VM and reusing that to spin CentOS 7.x builds to make gold images from it 17:57:46 mattdm: 18th right? 17:57:50 but it doesn't make sense for Fedora :( 17:57:57 spstarr: gotchya 17:57:58 jzb: yeah 17:58:05 OK 17:58:06 spstarr: I misunderstood, apologies 17:58:11 mattdm: do you have admin privs for Trac? 17:58:11 anyway since we're getting close to the hour.... 17:58:12 no worries 17:58:14 here's my thing... 17:58:15 https://apps.fedoraproject.org/autocloud/jobs/ 17:58:15 spstarr: that's ok. we're glad you're around the space and might have input for us in the future 17:58:22 I'm concerned that everything is purple 17:58:31 dustymabe: absolutely 17:58:50 I know kushal is unavailable right now, so I don't know who can figure out what's going on 17:59:02 mattdm: rtnpro might 17:59:07 also, the things that aren't purple are red 17:59:14 but he just left also 17:59:31 jzb: yes I have trac admin 17:59:41 ok I think we are done with meeting 17:59:45 * dustymabe sets fuse 17:59:49 3.. 17:59:52 mattdm: can you bump me up so I can add milestones? 17:59:55 ..2 18:00:01 ....1 18:00:09 #endmeeting