18:00:21 #startmeeting Infrastructure (2017-03-23) 18:00:21 Meeting started Thu Mar 23 18:00:21 2017 UTC. The chair is bowlofeggs. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:21 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:21 The meeting name has been set to 'infrastructure_(2017-03-23)' 18:00:23 #meetingname infrastructure 18:00:23 The meeting name has been set to 'infrastructure' 18:00:24 #topic aloha 18:00:26 #chair smooge relrod nirik abadger1999 lmacken dgilmore threebean pingou puiterwijk pbrobinson 18:00:26 Current chairs: abadger1999 bowlofeggs dgilmore lmacken nirik pbrobinson pingou puiterwijk relrod smooge threebean 18:00:27 #topic New folks introductions 18:00:43 oh whoops 18:00:48 #topic aloha 18:00:54 .hello puiterwijk 18:00:55 puiterwijk: puiterwijk 'Patrick "マルタインアンドレアス" Uiterwijk' 18:01:03 .himynameis cverna 18:01:04 cverna: cverna 'Slim Shady' 18:01:09 * yudocaa is here 18:01:13 morning everyone. 18:01:15 .hello sayanchowdhury 18:01:16 yudocaa: sayanchowdhury 'Sayan Chowdhury' 18:01:22 * nirik is in the no-go meeting, but will try and pay attention too. 18:01:35 .hello batman 18:01:36 vivek_: batman 'None' 18:03:01 #topic New folks introductions 18:03:02 * pingou here 18:03:21 is anyone new to fedora or the infrastructure group who would like to say hello? 18:03:33 Hello everybody! 18:03:40 hello cyberworm54! 18:03:59 hello 18:04:31 is there anything else that I've to do to get access to the aprentice group? 18:04:45 hola amigos 18:04:55 I've done pretty much everything from the getting started guide. 18:04:55 cyberworm54: are you more interested in sysadmin stuff or application development? 18:05:03 puiterwijk: sysadmin 18:05:13 And for apprentice group, please see me in #fedora-admin after the meeting 18:05:29 welcome cyberworm54! 18:05:33 puiterwijk: will do, thank you! 18:05:43 is anyone else here new who would like to introduce themselves? 18:05:45 .hello jcline 18:05:47 jcline: jcline 'Jeremy Cline' 18:06:11 ok, we will move on to announcements 18:06:20 .fas linuxmodder 18:06:26 #topic announcements and information 18:06:26 linuxmodder: linuxmodder 'Corey W Sheldon' 18:06:27 #info Fedora 26 Alpha freeze is on now! - everyone 18:06:29 #info Bodhi 2.5.0 beta deployed to staging, release notes available at https://bodhi.stg.fedoraproject.org/docs/release_notes.html#id1 - bowlofeggs 18:06:30 #info Fedora 26 alpha testing in full swing - kevin 18:06:32 #info distgit testcases enabled in production - tflink 18:06:34 #info There have been multiple problems with copr reported. 18:06:35 #info We will be moving stg hosts to new ip space after freeze. #5931 18:07:03 any other announcements, or shall we move on to discussion topics? 18:07:29 * jcline has nothing to report. All is quiet. 18:07:52 i'll add a detail about the bodhi 2.5.0 beta - the web ui has a new design by ryanlerch that looks really great. check it out at https://bodhi.stg.fedoraproject.org 18:08:09 not exactly infra related but we will have a influx of summer-applicants in the coming weeks and soem havce come in -admin for things for those not familiar pls direct them to -summer-coding 18:08:29 yeah, the new bodhi look in stg is pretty slick 18:08:58 bowlofeggs: I wonder if we could make the test tab in red or show a number for all failed tasks? something to bring attention if there's failures? 18:08:59 #info summer interns can be directed to #fedora-summer-coding 18:09:40 nirik, will have to look not spending as much time lately on anything stg.* 18:09:47 nirik: actually ryanlerch's newest PR does show a count of tasks, but doesn't differentiate success vs. fail - want ot make that suggestion there? https://github.com/fedora-infra/bodhi/pull/1379 18:10:02 i agree - it'd be nice if it showed a fail count 18:10:20 ok, let's move on to discussion 18:10:35 #topic OpenShift trial - kevin 18:10:48 nirik: ^ 18:10:57 * linuxmodder is 2 mtgs atm and making repo edits for respins sig if need me and no reply ping pls 18:11:11 ok... so I wanted to mention this... 18:11:20 we don't need to do much today aside from ponder on it. ;) 18:11:46 For a number of reasons I've been thinking about standing up a openshift instance in our infrastructure. 18:12:10 It's new and hot and all that, but also would help deploy some of these new services coming along (the factory 2.0 stuff, etc) 18:12:30 We are looking at a FAD to explore this more. 18:12:56 that would be in May? 18:13:02 There's no way it would replace everything we have now at all... but it could be along side it and a good fit for lots of things. 18:13:23 i spent some time using openshift in february and it did seem pretty nice 18:13:43 t would be radically different from how we do things now. 18:13:46 i've heard that it can be difficult to upgrade it from release to release, but i don't have any personal experience trying to do that 18:13:59 well, the ideal setup is: 18:14:05 I'm open to new and exciting things 18:14:44 you commit some (hopefully tested) PR, openshift sees this, builds a new container, builds your git checkout of your app, tests it, deploys it and lets it have some traffic, then gradually replaces all the old ones with this new one. done. 18:14:59 so the idea is that you are always deploying 18:15:20 I have tested building flask app from pagure with openshift :) 18:15:20 very cool and useful sounding.. 18:15:21 And if a commit doesn't start for whatever reason, it will automatically roll it back 18:15:24 and that your CI and will test everything 18:15:36 yeah. 18:15:53 So in this world... you wouldn't need staging. Or have freezes. 18:16:08 so you need to work on the CI part as it could break but if there aren't tests for that breakage 18:16:14 but also some things don't fit well in openshift IMHO... like databases. Or proxies. 18:16:15 s/you/the developer/ 18:16:19 or builders 18:16:29 smooge: yep. absoletely 18:16:49 how would the push to prod work tho or is it not really a stg ?prod thing with openshift ? still not spun up on it much 18:16:52 yeah database schema migrations would not work well with the 'roll it back if it doesn't work' part 18:17:14 linuxmodder: theer would be no need for staging... 18:17:19 ah 18:17:30 so basically dev then prd? 18:17:32 you test your app by testing it with jenkins or whatever then deploying it and making sure it runs 18:17:33 bowlofeggs: that's why you need to make schema migrations very rare :) 18:17:44 haha 18:17:53 bowlofeggs: not even a joke. 18:17:58 just a wish 18:18:01 schema changes should be rare 18:18:02 it is possible to use it without connecting it so directly to git 18:18:03 bowlofeggs, just use mongo, no schema means no migrations means no problem 18:18:08 hahaha oh my 18:18:09 *should* being a keyword :) 18:18:24 bowlofeggs: well, git is pretty tied in, but you can do just containers... 18:18:36 ie, a container that installs rpms or whatever 18:19:03 yeah that's what i mean 18:19:16 i don't agree that schema migrations should be rare - they happen when they need to happen 18:19:17 right we could skip the CD part if desired. 18:19:39 but it has advantages too... saves all the work making a release, building a rpm, signing, etc. 18:19:46 not all new features can be accomplished without migrations, and features priorities are not always decided by us 18:19:53 bowlofeggs: in an ideal world they should, in practice they don't 18:19:58 * jcline has two pending schema migrations for anitya this release alone 18:20:24 in theory, theory and practice are the same. In practice they are not. 18:20:29 bowlofeggs: sure. But most changes can be made backwards compatible. Aka, adding a new table wouldn't be a problem when you roll back 18:20:40 Anyway, let's keep that discussion for some othe time 18:21:12 also, another question around this is how much we want ansible to manage things in a openshift world 18:21:33 since it has it's own command line and web interface... do we make ansible configure all apps and not use those? 18:21:53 I would say we make ansible configure it all, yes 18:21:55 again, nothing needs deciding today, just lots to think about 18:22:15 puiterwijk: well, you mean deploy it, but also control say how many pods on app foo? or ? 18:22:24 nirik: also deploy the apps itself. 18:22:42 ok. 18:22:43 How many pods - maybe not, but creating the image streams, routes and deployments: yes 18:23:03 I know origin has deployment ansible scripts for everything... 18:23:09 (for the base setup) 18:23:26 That's also the deployment mechanism for the RH product 18:23:32 I'm still trying to understand how it all works and fits too. :) it's really complex. 18:24:13 Things like koji I don't see fitting in this at all... 18:24:28 but things like fedocal or whatever could fit nicely 18:24:40 copr :) 18:25:29 gives himself a ಠ_ಠ, and moves back to topic 18:25:31 The frontend sure... the backend, perhaps, the builders... not sure at all. Containers don't contain enough for me. 18:25:43 * nirik looks for the fad page 18:25:44 yeah i think it'd be good to pick a very simple app to test with first and see how it goes 18:25:59 How about Ipsilon? :) 18:26:12 https://fedoraproject.org/wiki/CI_and_Infrastructure_Hackathon_2017 18:26:31 (I actually have a production deployment of Ipsilon running on Openshift. It works nicely) 18:26:33 all tenative so far... but if you are in RDU area or can get there. ;) 18:26:57 how about download servers? ;) 18:27:46 sounds cool to me. any other thoughts, or shall we move on? 18:27:50 anyhow, just wanted to bring it up so everyone thinks about it... I figure we might be able to scrape up 2 machines to setup a test instance on 18:27:51 I'd be down for that hackathon but may have to do remote 18:28:21 feel free to move on 18:28:42 #topic Apprentice Open office hours 18:29:16 do any apprentices have any questions? need help finding things to do? 18:29:57 Any apprentice workday planned? 18:30:02 https://fedoraproject.org/wiki/Workday_planning_page 18:30:27 I am looking for some ansible stuff 18:30:52 There's currently none planned, I think. Feel free to suggest one on the list 18:31:13 well stuff meaning if I can help with anyhting.. 18:31:37 maybe if anyone who has time can toss me an assignment or the like 18:31:54 capitanocrunch, ForrestalGrim I would think an apprentice work day could be done after F26 release 18:32:11 we could do one before hand in a non-freeze period 18:32:20 ForrestalGrim, tried looking onteh easyfix page for some ansible ? 18:32:20 ForrestalGrim: did you look at https://pagure.io/fedora-infrastructure/issues?status=Open&tags=easyfix if there's anything interesting for you there? 18:32:25 Ok. 18:32:29 Not fully understanding the process, Ill take whatever :) 18:32:35 once I get added, I am going to start looking at the easyfix and start working on somehting 18:32:36 i've got an ansible thing i've been meaning to do actually - bodhi uses two separate files for its staging.ini and production.ini templates - it would be better if there was just one settings template and the staging/production stuff was just if statements in it 18:32:50 capitanocrunch, I would say the apprentice day would be good for you to bring up on the list 18:32:51 smooge, possible to do one in between beta and rc ? 18:32:55 i haven't filed an issue abotu that, but if it sounds fun you could try that out 18:33:07 insofar as the easyfix not I need to inspect those more. I was playing the assumption game than advanced stuff like that wouldnt be easyfix. Maybe my bad there 18:33:08 and show some of the pre flip and such 18:33:14 I'll look deeper tonight 18:33:44 FYI, the f26 alpha go/no-go meeting just finished and we are no go. So that means another week of freeze. :) 18:33:58 it's getting too cold around here 18:33:59 so I might ask for a FBR to upgrade pagure 18:34:04 we could do another docs work day during freeze (which would be fine since docs aren't frozen) 18:34:14 pingou: feel free. :) 18:34:14 2.14 brings some IO optimization which I think we be nice to put out 18:34:19 worst it could open other doors with things you may not have seen as someone wants your ideas / help elsewhere from work on an easyfix ForrestalGrim 18:34:22 the changelog is rather long though :( 18:34:36 linuxmodder very true 18:35:01 bowlofeggs do you have a ticket on your bohdi/ansible thing? 18:35:06 somehtng I can read? 18:35:22 ForrestalGrim: i don't, but i can file one and send you a link after this meeting 18:35:35 bowlofeggs thanks, thatd be much appreciated 18:35:37 it was just something in my soft, human brain 18:35:41 lol 18:36:54 any other thoughts? 18:36:57 questions? 18:37:54 #action bowlofeggs will file a ticket about unifying bodhi's staging/production.ini files for ForrestalGrim to check out 18:38:06 ok, let's move on to open floor 18:38:12 #topic Open Floor 18:38:56 18:39:13 :) 18:39:49 Any other questions or remarks? Favorite build systems? 18:40:07 imake! 18:40:08 Fedora! 18:40:09 * nirik runs 18:40:22 * bowlofeggs installs koji on puiterwijk's smart watch 18:40:43 * puiterwijk is surprised nirik and bowlofeggs didn't mention COPR just to mess with me :D 18:41:00 bowlofeggs: which smart watch? I can't even wear a normal watch :-). But I do have koji run everywhere! 18:41:14 puiterwijk, I did but was promptly ignored 18:41:43 thanks guys! (boss just walked in) 18:42:09 hehehe 18:42:23 cool, well it's about that time 18:42:29 #endmeeting