14:00:07 #startmeeting Infrastructure (2018-09-20) 14:00:07 Meeting started Thu Sep 20 14:00:07 2018 UTC. 14:00:07 This meeting is logged and archived in a public location. 14:00:07 The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:07 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:00:07 The meeting name has been set to 'infrastructure_(2018-09-20)' 14:00:07 #meetingname infrastructure 14:00:07 The meeting name has been set to 'infrastructure' 14:00:07 #topic aloha 14:00:07 #chair nirik pingou puiterwijk relrod smooge tflink threebean cverna 14:00:07 Current chairs: cverna nirik pingou puiterwijk relrod smooge tflink threebean 14:00:17 morning. 14:00:24 good morning 14:00:40 good morning all 14:00:41 evening :-) 14:01:07 mkonecny: good afternoon :D 14:01:17 .hello2 14:01:18 bowlofeggs: bowlofeggs 'Randy Barlow' 14:01:32 #topic New folks introductions 14:01:32 #info This is a place where people who are interested in Fedora Infrastructure can introduce themselves 14:02:44 hello anyone new this week? 14:03:04 each day is like it's new again! :) 14:04:04 #topic announcements and information 14:04:04 #info tflink is on extended PTO 14:04:04 #info Beta Freeze is in effect. All changes to frozen systems in Infrastructure will require +1 14:04:09 * aikidouke thinks nirik may be using my brain or has a similar model 14:04:14 There weren't any other announcements 14:04:24 ha. 14:04:45 yeah, it's been pretty quiet freeze week 14:05:25 #topic Oncall 14:05:25 #info Smooge is on call from 2018-09-13->2018-09-20 14:05:25 #info Nirik is on call from 2018-09-20->2018-09-27 14:05:26 #info ??? is on call from 2018-09-27->2018-10-04 14:05:26 #info Summary of last week: (from Smooge) 14:05:31 #info bodhi 3.10.0 is released and deployed to staging - waiting for freeze to end 14:05:42 thanks bowlofeggs 14:05:44 fall is coming! 14:06:07 oh yes! ^ and autumn as well 14:06:37 The biggest problem was reg-server on sundries went crazy on the weekend 14:06:47 oh yeah, we still need to debug that? 14:06:59 It needs a newer version installed which will look at multi-arch 14:07:21 beyond that it was mostly pings 14:07:28 I think we have that in staging 14:07:35 I need to have someone to take the week after Nirik and the week afterwords 14:07:43 multi-arch reg-server 14:08:12 So 2018-10-04 -> 2018-10-11 and 2018-10-11 -> 2018-10-18 14:08:29 cverna, I don't know if the rpm was done 14:08:38 cverna: oh? I was working on the new rpm, but it wasn't finished. 14:08:40 smooge: feel free to give me one of the slots. 14:08:49 and puiterwijk was going to look, but he's been busy 14:09:07 puiterwijk, 2018-10-04 sound good? 14:09:11 smooge: sure 14:09:13 nirik: smooge : ok I though relrod did that couple weeks ago but I might be wrong 14:09:19 thought* 14:09:41 oh foo I got the dates wrong 14:09:52 2018-09-27->2018-10-04 14:10:03 does that still work or do you need the next week puiterwijk ? 14:10:13 smooge: sure 14:10:45 ok thanks 14:10:49 I have updated the gobby 14:11:17 #topic container registry and Quay.io - cverna 14:11:30 I take it over :) 14:12:05 So yesterday I has some update from the Red Hat OSBS team and they are preparing to move to Quay.io 14:12:28 Multi-arch and OCI support is expected to be available in October 14:12:40 cverna: is it open source yet? 14:12:45 huh, any chance there will be koji integration? ;) 14:13:03 I wanted to get the general feeling about moving Fedora's registry to Quay.io and not running the registry in our Infra 14:13:24 cverna: if it's FOSS: let's move ASAP. If not, wait until it is 14:13:40 .hello2 14:13:41 x3mboy: x3mboy 'Eduard Lucena' 14:13:43 puiterwijk: +1 14:13:55 It is not yet FOSS but as far I as I know docker hub is not FOSS 14:14:11 nirik: the koji integration is managed by OSBS 14:14:20 cverna: we run https://github.com/docker/distribution 14:14:23 Which is FOSS 14:14:25 oh, I see, this is the registry... misread 14:14:28 * otaylor notes that regindexer will still be necessary at least in the near-to-medium turn, but should be possible to adapt to run against quay.io. 14:14:39 puiterwijk: with Quay we would not run anything 14:14:51 puiterwijk: we would just use Quay.io as a service 14:15:02 cverna: sure. But it's still not FOSS. Even for services, we strongly prefer/require FOSS 14:15:24 cverna, there has been a lot of pushback from our community and at times the council to keep services we use FOSS 14:15:35 Ok that's why I brought up the topic, I wanted to get a general feeling 14:15:41 the github usage has been a thorn 14:15:50 Right. So the general feeling is: if it's FOSS: YES. If not: wait. 14:15:57 note that greenwave/waiverdb is already using quay 14:16:01 yes github and docker hub are definitely not FOSS 14:16:22 pingou: right. But there's a difference between us grabbing FOSS contents from a non-FOSS registry versus sending all our users there to download the official images 14:16:29 the s2i image we use to build silverblue and coreos image is from quay.io too 14:16:32 fair 14:16:37 cverna: see my note 14:17:50 ok then we can stick with our current solution until Quay become FOSS 14:17:54 is there plans to opensource quay? I would think so, but haven't seen anything specific 14:18:12 I expect it is similar to getting tower opensourced 14:18:16 nirik: well, usually that's not shared until the message is "It's FOSS as of today", right? 14:18:34 As I understand the quay team is one person 14:18:48 so I think getting it FOSS is not priority currently 14:18:49 wow, really? 14:19:10 they are looking to expand but yeah 14:19:16 note also that our registry hasn't required very much maintenance/upkeep 14:19:19 it will end up FOSS 14:19:23 but as usual no ETA 14:19:24 so it's not costing us too much 14:19:37 bowlofeggs: our user experience to the registry is horrible 14:19:48 https://registry.stg.fedoraproject.org/repo/fedora/tags 14:20:03 you can't even access which tags are available 14:20:28 the new reg fixes that. 14:20:36 but yeah, it's been a pain 14:20:40 fair 14:20:55 i was thinking more about the behind the scenes thing than the UI 14:21:09 * bowlofeggs hasn't ever touched the UI thing 14:21:33 ok, so anyhow that was my head ups on this topic :) 14:22:18 perhaps we also want to think of uploading to quay.io even if we don't advertise it as the main place? 14:22:30 or is that too much work? 14:22:52 we could do for base images 14:23:07 like we are doing on dockerhub 14:23:40 i'd be +1 to also uploading to quay, since that's similar to what we do with dockerhub 14:23:50 yeah... then if we switch we are already ready and people won't see nothing there in the mean time 14:24:06 it would also allow us to claim namespace 14:24:14 yeah 14:24:40 OK I ll open a releng ticket for that, and will look at what is needed 14:25:17 i could expand bodhi's container support to allow publishing to a list of registries 14:25:27 it currently wants to publish to exactly one 14:25:34 unless we only care about the base image for now 14:25:45 i also have like 0 time to work on that :( 14:25:51 might be nice, but when one was down would that fail the push? 14:26:01 nirik: can of worms! 14:26:02 hahah 14:26:05 yeah. 14:26:10 bowlofeggs, bodhi is completely covered with tests so its done right? so you should have all the time in the world 14:26:16 yeah i dunno what would be the right behavior there 14:26:23 smooge: lol 14:26:28 smooge: :D 14:26:34 smooge: have you ever considered doing stand up? 14:26:48 bowlofeggs: let's do at least one a day! 14:26:55 from a workflow standpoint it seems like uploading to ours and then having something sync from there to other places would make sense. 14:26:56 hahaha 14:27:03 nirik: yeah that's true 14:27:10 that would be simpler 14:27:26 skopeo on a cron 14:27:46 we could just let jcline sit around all day with skopeo and run it by hand 14:27:58 Sounds good to me 14:28:39 * nirik still thinks skopeo sounds like a 70's detective show 14:29:00 It also depends how the registries work, for example dockerhub process requires us to push the images in a git repo :S 14:29:14 cverna: yeah that one is super weird 14:29:17 yeah, thats not ideal. ;( 14:29:27 I have opened https://pagure.io/releng/issue/7828 14:29:27 * nirik wonders if we could push a redirect. ;) 14:29:36 nirik: nope. 14:29:43 bummer 14:29:51 ok do we have much more on this topic at the moment? 14:30:07 Not on my side :) thanks all for the feedbacks 14:30:24 thanks for bringing it up cverna. ;) 14:30:25 Scopeo.. A Martin Quinn production! 14:30:49 #topic Tickets discussion 14:30:50 #info https://pagure.io/fedora-infrastructure/report/Meetings%20ticket 14:31:09 not sure I updated this this week. 14:31:19 oh, perhaps I did. 14:31:28 I thought those were last wekes 14:31:42 yeah, did we get to them last week tho? 14:32:01 yep 14:32:08 threebean did a long talk on it 14:32:09 ok, then moving on is fine. ;) 14:32:23 so what I would like to do is ask if we could add another item 14:32:37 should we move towards no unassigned tickets 14:33:05 hum, not sure... the problem is that there are few of us and lots of tickets... and people might want to take something someone else has, etc. 14:33:09 we have some set of tickets which have been unassigned for a long time 14:33:45 so maybe we could go over which ones have been that way for a long time and just try to get one either assigned or closed per week? 14:33:48 yeah, ideally we would finish//close the ones we have assigned and take more 14:34:15 I took a couple of tickets this week that way myself. 14:34:38 but the long term ones I am not sure anyone can work on without some discussion 14:34:53 a script that asks the owner if they still need $thing fixed and, sleeps for 60 and then closes the issue? 14:35:09 I don't think one size fits all. 14:35:17 * aikidouke needs more coffee 14:35:19 yeah.. 14:35:20 but I'm fine discussing old tickets in meeting 14:35:36 but the answer on some might be 'we still want to do this, no one is working on it, etc' 14:35:46 well it was like most of my ideas at the spur of the moment... 14:36:04 #topic Apprentice Open office minutes 14:36:04 #info A time where apprentices may ask for help or look at problems. 14:36:05 smooge: What are some examples of the longer-term ones? Could we reasonably knock a few of those out over a day or two vFAD? 14:36:40 relrod: look at the last page of tickets... 14:36:41 relrod https://pagure.io/fedora-infrastructure/issue/6750 14:36:49 Hi 14:37:11 vFAD is? virtual Fedora Apprentice Day? 14:37:13 that ol chestnut 14:37:23 aikidouke: activity day* 14:37:23 Virtual Fedora Activity Day 14:37:25 https://pagure.io/fedora-infrastructure/issue/5943 14:37:31 :) ty 14:37:55 Can we look at issue 7105? 14:38:06 I'm not sure if it's still relevant 14:39:39 .ticket 7105 14:39:42 smooge: Issue #7105: sudo is not working on f28-test.fedorainfracloud.org - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/7105 14:40:10 hum. yeah I think thats fixed now 14:40:21 after doing updates and poking at it. 14:40:23 thanks for pointing that out x3mboy 14:40:23 Ok :( 14:40:24 * nirik looks to see 14:40:34 I don't know for sure... 14:40:47 I will look for another ticket to work on 14:41:14 yeah, looks like it's working... :( 14:42:55 sorry x3mboy, I should have updated that when I saw it working the other day. 14:43:44 we need some more easyfix tickets. ;) 14:44:22 Yeah! More easyfix!!! 14:44:39 #topic Need more easyfix 14:46:38 sorry that should have been info 14:46:40 three's a handful of bodhi easyfixes (python) 14:46:42 #topic Open Floor 14:46:55 which fits also with open floor 14:46:55 I'll try and think of some... but not sure off the top of my tired brain what they might be. ;) 14:47:35 https://github.com/fedora-infra/bodhi/issues?q=is%3Aopen+is%3Aissue+label%3AEasyFix 14:47:38 (currently 14) 14:48:07 On Pagure the most recent easyfix is 10 month old 14:48:16 I have plenty of easyfixes on Anitya https://github.com/release-monitoring/anitya/issues?q=is%3Aissue+is%3Aopen+label%3Aeasyfix 14:48:43 x3mboy: wrong :) 14:48:58 * pingou added at least one earlier this week 14:49:03 I think we need to add pointers to what needs to be done in our easyfix otherwise they might not be that easy :) 14:49:10 x3mboy: you mean the fedora-infrastructre pagure instance? 14:49:17 nirik, yes 14:49:25 I stand corrected then :) 14:49:57 easyfixes are hard... a thing thats easy to do, but you can live without it being done for a while even tho it might be a few minutes to fix 14:50:16 yeah 14:50:20 yes :S 14:50:36 but they are important to get new folks up to speed on how things are setup... :) 14:50:41 For me, the main problem is where the issues are displayed 14:50:42 cverna: I'm adding task list to each easyfix 14:50:52 mkonecny: cool :) 14:50:56 e.g.: I always look on Pagure 14:51:10 Now I'm seeing several githubs 14:51:38 x3mboy: Look here https://fedoraproject.org/easyfix/ 14:51:48 x3mboy: if you are looking for sysadmin tasks the pagure repo is the right place to look 14:52:10 cverna, yes, that's what I'm looking for 14:52:12 also, github apparently supports new labels for this officially 14:52:18 and they aren't the same as what bodhi uses 14:52:27 but if i change bodhi it will break that easyfix app 14:52:28 haha 14:52:47 of course 14:52:47 bowlofeggs: I also noticed this 14:52:53 supposedly you can use "help wanted" or "good first issue" and github will direct newcomers to those in some way 14:53:06 Oh tagger 14:53:09 I will miss you 14:53:16 i wish i could just click a bool on my EasyFix label in github and mark it as doing that for bodhi 14:53:23 that would be better than magic string 14:53:24 s 14:53:26 but oh well 14:53:30 we can use both on issues 14:53:39 ok I need to bring this meeting to a close soon 14:53:46 i might start using it and just send a PR to the easyfix app 14:53:54 but no idea when such a patch could get deployed 14:54:39 OMG I'm the contact for easyfix in mindshare 14:54:43 :P 14:55:32 ok lets move this to the list and we can see how we are next week 14:55:41 Thank you all for coming this week 14:55:44 #endmeeting