14:00:00 #startmeeting Infrastructure (2018-04-26) 14:00:00 Meeting started Thu Apr 26 14:00:00 2018 UTC. The chair is relrod. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:00 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:00:00 The meeting name has been set to 'infrastructure_(2018-04-26)' 14:00:00 #meetingname infrastructure 14:00:00 The meeting name has been set to 'infrastructure' 14:00:00 #topic aloha 14:00:01 #chair smooge relrod nirik pingou puiterwijk tflink 14:00:01 Current chairs: nirik pingou puiterwijk relrod smooge tflink 14:00:11 .hello2 14:00:12 o/ 14:00:14 Good morning, everyone! 14:00:18 abompard: abompard 'Aurelien Bompard' 14:00:20 good morning folks. 14:00:21 .hello2 14:00:26 creaked: creaked 'Will Chellman' 14:00:33 hello 14:00:35 hey folks! Thanks for waking up so early :-) 14:00:55 hi o/ 14:00:57 * pingou agrees with abompard and sends coffee to nirik 14:00:59 .hello sinnykumari 14:01:00 ksinny: sinnykumari 'Sinny Kumari' 14:01:15 morning 14:01:22 abompard: no problem. Glad to have the .fr contingent here. ;) 14:01:30 hello! 14:01:34 :+1: 14:01:39 nirik: Evolution may not agree with this :D 14:01:46 something to do with app names iirc 14:02:00 .hello2 14:02:01 dustymabe: dustymabe 'Dusty Mabe' 14:02:44 Welcome all, nice to see everyone :) 14:03:02 #topic New folks introductions 14:03:02 #info This is a place where people who are interested in Fedora Infrastructure can introduce themselves 14:03:18 Do we have any newcomers who would like to give an introduction? 14:03:54 I'm new here. I don't have much of an introduction but excited to help out! 14:04:18 welcome creaked. :) 14:04:24 welcome creaked 14:04:29 do feel free to ask questions as we go... 14:04:59 cverna: are you interested in the apps development side of things, or sysadmin, or? 14:05:03 creaked: * 14:05:43 * relrod notes to self: c doesn't work when it sorts by last-spoke and someone else talked ;) 14:05:43 * cverna is interested in everything :) 14:05:45 both. Trying to get a feel for the infrastructure currently 14:05:45 Hi all! I am Sinny Kumari working in Fedora Engineering team. I work on Fedora Atomic and getting familiar with Fedora infrastructure with the help of folks available here. I am excited to learn and help more around our Fedora Infrastructure :) 14:05:56 ksinny++ 14:06:27 creaked: cool. Well welcome, nice to have you 14:06:33 ksinny: welcome :) 14:06:42 .hello2 14:06:43 bowlofeggs: bowlofeggs 'Randy Barlow' 14:07:02 Alright, well let's move on to announcements for the week 14:07:04 Thanks everyone for shifting the meting time :) 14:07:04 meeting* 14:07:22 ksinny: yeah this seems to be working well, nice turn-out this week 14:07:33 #topic announcements and information 14:07:33 #info Fedora 28 Final freeze is in effect! 14:07:33 #info new priority fields in infrastructure issues, please use them! 14:07:33 #info Bodhi 3.6.1 seems to be stable in OpenShift since Tuesday 14:07:33 #info ask is being moved to new host 14:07:41 #info: pagure 4.0 and 4.0.1 released 14:07:54 (and running in stg) 14:08:36 pingou: oh, does that mean that you're interested in merging my Py3 branch? 14:08:50 abompard: def :) 14:08:56 cool, will rebase 14:09:00 \o/ 14:09:07 bowlofeggs: is next a 3.6.2 bodhi with that waiving/tests showing? 14:09:15 abompard: I'd like to wait for 4.0+ to make it to prod, in case we need more .z, but py3 will def be in 4.1 :) 14:09:32 pingou: when did you want to go to prod? ;) 14:09:45 well... :D 14:10:04 how on time is F28 looking? 14:10:25 nirik: yeah i'm planning a 3.6.2 to be next 14:10:29 go/nogo is later this morning... and so far it's looking pretty goish... 14:10:44 but you never know what will show up at the last minute. ;) 14:10:48 haha yeah 14:10:59 i did consider whether to just make a 3.7.0 if we are go 14:11:15 but it would be faster to just do a 3.6.2 and i know pingou wants those patches :) 14:11:15 if we're go, then I'd say let's wait 14:11:27 bowlofeggs: I do :] 14:11:32 pingou: well i can deliver a 3.6.2 faster still since it's a smaller diff 14:11:41 a 3.7.0 would make me want more testing time 14:11:45 so i think 3.6.2 is ideal 14:11:54 just due to the number? ;) 14:11:56 thanks :) 14:12:10 nirik: well 3.7.0 means feature patches, 3.6.2 means just 4 smallish patches 14:12:27 if we're no go, I'd be considering upgrading pagure.io to 4.0 14:12:30 i.e., the difference is the amount of change in the diff and the purpose of the release 14:12:46 with the tasks split over multiple queues, it should help the slowness in the UI we're seeing these days 14:12:48 nirik: i follow https://semver.org 14:13:25 bowlofeggs: right, just the way it sounded there was there were no different commits, just the number... but right... 14:13:44 pingou: if we are go, just wait for next week then? 14:13:48 yes 14:13:52 nirik: ah no - the number is very meaningful in bodhi due to semver 14:13:56 if we're nogo, let's discuss :) 14:14:00 i.e, not marketing nonsense :) 14:14:15 linux 5.0! 14:14:19 haha yeah 14:14:20 :D 14:14:25 bowlofeggs: skip the 3 and go to Bodhi 7! 14:14:31 i wish everyone did semver in the whole universe 14:14:33 Bodhi X? 14:14:34 haha 14:14:38 Alright, any other announcements? 14:15:00 Bodhi 7 with a react front end :) 14:15:11 Reminder that we exit freeze the day after the release 14:15:30 whenever that may be :) 14:15:58 crystal ball says the release will be on a tuesday 14:16:20 good bet... but what year? ;) anyhow... move on... 14:16:28 Alright, so let's go through some tickets 14:16:32 nirik ▸ ;) 14:16:36 #topic Tickets discussion 14:16:44 #link https://pagure.io/fedora-infrastructure/issues?status=Open&priority=2 14:17:05 so, these are a few I wanted to bring up in the meeting. 14:17:10 tickets! get your tickets! 14:17:11 shall we do them one at a time? 14:17:17 I guess we'll just go down the list in order? 14:17:20 #link https://pagure.io/fedora-infrastructure/issue/6750 14:17:27 BTW, I think there's a typo in the "Waiting on Asignee" pagure priority. There's an "s" missing. 14:17:41 .ticket 6750 14:17:43 puiterwijk: Issue #6750: Help! We need some kind of search solution for docs - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/6750 14:17:46 relrod: ^ 14:18:13 so, this got deferred from the hackfest. Do we have any kind of plan aside that? 14:18:14 puiterwijk: But #link makes it pretty in the html, I think! :P 14:18:19 iirc 14:18:32 relrod: just the link in IRC does that too :) 14:18:40 And this way, we get more background info 14:18:44 okay then 14:18:46 * relrod shuts up 14:18:50 I like this meeting. So organized 😀 14:19:06 * nirik sighs. fas 500'ed to ipsilon again... 14:19:29 nirik: I think someone was going to get in touch with people on the info? 14:19:35 nirik: huh... :( 14:19:44 Maybe we could look at postgres full text search for the documentation 14:19:56 cverna: we don't have the docs in a database. 14:19:58 abompard: fixed the missing "s" 14:20:04 They're static content 14:20:06 relrod++ 14:20:06 abompard: Karma for codeblock changed to 8 (for the f27 release cycle): https://badges.fedoraproject.org/tags/cookie/any 14:20:19 I don't have to have the full elasticsearch stack 14:20:25 we have gone thru this N times sadly... 14:20:37 the open source solutions seem either lacking or... java. 14:20:40 puiterwijk: we could index the documentation 14:20:49 cverna: right, but as I already told people, if we want to, we can just outsource the elastic part... 14:20:57 i've heard good things about elastic search before, but i've not used it first hand 14:21:20 cverna: right, but then we need someone to write and maintain a new codebase for it... 14:21:39 outsourcing sounds like a good plan 14:21:40 While search is a well-researched topic, with lots of implementations. Some FOSS ones that are available hosted. 14:22:07 ok +1 for outsourcing 14:22:14 so, we don't need to solve this here... I just wanted everyone to think about it and perhaps we discuss on list? and/or go back to standing up a POC elastic 14:22:21 * puiterwijk points at https://www.elastic.co/cloud - managed by the original authors 14:23:05 nirik: btw I asked you one day to spin up a fedorainfracloud machine to test elasticsearch, I won't have the time to work on it so feel free to reclaim the resources 14:23:13 .hello2 14:23:14 x3mboy: x3mboy 'Eduard Lucena' 14:23:29 abompard: ok. Will look for that later... thanks for letting me know. 14:23:34 * x3mboy just reading the meeting 14:23:37 I think outsourcing it is a good idea, but also (re a lot of the FOSS ones being Java) - we do have openshift now, so maintaining the Java mess might be slightly less bad (or at least slightly more contained) 14:23:56 puiterwijk: so, that would be us spinning up their stuff in aws and running it to index our content? or just paying them to do all of it? 14:24:09 nirik: second one. 14:24:27 relrod: possibly... 14:24:28 You just click "3 nodes", and they give you an username and API key 14:24:39 relrod: you still don't escape the xml 14:24:42 o lord, the xml 14:24:48 relrod: that'd still need lots of storage and performance tuning... 14:25:05 yeah 14:25:12 Just tossing it out as something to keep in mind 14:25:17 Not advocating for it 14:25:19 java is a great language choice if you have a lot of xml and need to turn it into very long backwards null-pointer tracebacks 14:25:39 hahaha 14:25:45 ok, so lets move this to list? Someone want to start a thread there on search? 14:25:51 nirik: to be specific, I meant https://www.elastic.co/cloud/as-a-service 14:25:58 or puiterwijk can just add his idea to the ticket I guess? 14:26:06 I have no idea if we have budget for that kind of thing 14:27:15 Well, I know we've got some people we know that work at Elastic. So I can ask people (or ask people to ask people) if they would like a link on our sponsors page or the like. 14:27:39 not sure if we can get by with their starter pack, but their cheapest service is $45/mo 14:27:42 But anyway, let's ask Matt what he thinks? 14:27:47 that's way cheaper than paying an engineer to solve it 14:27:50 if they are willing to comp us that might be cool. I don't really think we have that much stuff to index next to say...ibm 14:28:05 puiterwijk: sure! 14:28:23 puiterwijk: can you update the ticket with that info? 14:28:27 nirik: will do 14:28:35 and set it 'waiting on reporter' 14:28:38 pingou: what am I not agreeing with? outsourcing ELK? 14:28:44 Evolution: frenchies 14:28:59 pingou: we know that already, no need to discuss 14:29:14 pingou: I don't trust anyone who eats that much cheese, you're correct 14:29:16 :-P 14:29:19 nirik: okay to move on to the next ticket? 14:29:23 sure. 14:29:30 .ticket 6563 14:29:30 Evolution: you're just being jealous :-p 14:29:32 relrod: Issue #6563: request for haskell-sig group in src.fedoraproject.org - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/6563 14:29:47 so, this is stalled out because we have a fas user called 'haskell-sig' 14:30:02 and fas won't let you make a group when the user exists. 14:30:12 any ideas for moving it forward? 14:30:27 pingou: I can't get good raclette here, so yes. yes I am 14:30:29 rename user? 14:30:32 call it haskell-cig? 14:30:35 is the user active? 14:30:46 the -sig is no longer a requirement for groups w/ pagure 14:30:59 pingou: rename user? or rename group? 14:31:06 nirik: haskell-sig-group? 14:31:11 so we could also just name the group differently, but the -sig makes a nice pattern 14:31:13 Renaming a user is... really hard 14:31:17 yeah 14:31:18 nirik: I was thinking rename the user 14:31:36 pingou: I think renaming a user is more problems then it's worth 14:31:40 ok 14:32:01 call it "haskell-sig'; DELETE FROM users" 14:32:08 hum, how about 'haskell-packaging-sig' ? 14:32:09 +1 to haskell-sig-group 14:32:15 nirik: +1 14:32:20 nirik: +1 14:32:30 abompard's one is tempting... 14:32:32 oh +1 too :) 14:32:52 I.... think it might be protected against that?... 14:32:54 * puiterwijk opens FAS code 14:33:35 ok then... pingou: you want to do the group as haskell-packaging-sig ? or you want me to? 14:33:37 nirik: that implies the SIG _only_ does packaging stuff...which is fine if that's the goal, but... 14:33:51 relrod: yeah, thats the danger there... 14:34:01 nirik: on it 14:34:04 how about haskell-lang-sig? 14:34:13 the group is for being POC for haskell packages 14:34:24 relrod: fine with me too. 14:34:36 pingou: ^ 14:34:45 abompard: that should be fine :P 14:34:52 or maybe fedora-haskell-sig, though not sure if we have a policy against 'fedora' in group names 14:35:00 petersen as group owner? 14:35:04 puiterwijk: haha :-D 14:35:08 I think haskell-lang-sig works good 14:35:13 bowlofeggs: validators.Regex(regex='^[a-z0-9\-_]+$'), 14:35:16 :D 14:35:18 works well 14:35:50 glasgow-haskell-sig ? ;) 14:36:09 anyhow, lang works for me... and yes peterson as owner 14:36:11 :D 14:36:24 okay moving on becasue I want to make sure we get through all of these 14:36:28 .ticket 6424 14:36:29 * nirik nods 14:36:29 relrod: Issue #6424: pagure-dist-git doesn't enforce groups to have "packager" prerequisite - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/6424 14:36:33 .members cla_done 14:36:38 abort abort 14:36:45 pingou: Members of cla_done: a126ium a15 a1pha a2batic a2n a3b3 a4asystems a4lg a69 a6hijeet a7ndrew a864800665 a9074 a9577095 a9rkzz aa7955 aaabbb0 aaabbb1 aaabbb2 aaabbbccc aaabbbccc1 aaabbbccc10 aaabbbccc2 aaabbbccc3 aaabbbccc4 aaabbbccc5 aaabbbccc6 aaabbbccc7 aaabbbccc8 aaabbbccc9 aaacccaaa aaacccaaa1 aaacccaaa11 aaacccaaa12 aaacccaaa2 aaacccaaa21 aaacccaaa4 aaaddd aaamourao aaasddsadsa aaasddsadsaklj aaasssddd1 (50 more messages) 14:36:47 .members haskell-lang-sig 14:36:50 pingou: Members of haskell-lang-sig: @petersen 14:37:06 oof 14:37:15 * nirik is glad of paging. :) 14:37:19 anyhow, on this one... 14:37:38 nirik: do you do the list? 14:37:44 or do they already have one? 14:38:33 they do not, I can make one... also haskell-lang-sig ? 14:39:37 wfm 14:40:05 and done 14:40:07 so pagure-dist-git and groups... 14:40:25 right, so on this... right now only packagers can ssh to anything on src.fedoraproject.org 14:40:44 ah so that would save us for now 14:40:49 eventually that will change and we can allow anyone we like 14:40:50 but we also wanted to cahnge that 14:41:07 (well, not ssh, but push via https) 14:42:16 nirik: but even with that, I think we'd want to make sure they're at least in the packager group for "real" packages, not? 14:42:25 (i.e. non-forks) 14:42:32 yes 14:42:45 I guess this ticket only happened because people were allowed to make groups? 14:42:59 which is disallowed now? 14:43:43 yup 14:44:06 it's also because people are currently allowed to fork, but can practically do nothing with their fork unless they are packager 14:44:22 so... closed -> fixed? or delete that group and close? or ? 14:45:05 closed -> notabug? 14:45:07 well, it is still valid, pagure doesn't enforce the packager group 14:45:45 would it be easy/doable to add that enforcement? 14:45:53 Oh, right... For which, we'd need to 1. make sure that all groups added to Pagure require packager, or 2. add code to Pagure to check packager group membership 14:46:23 and then what do we do when we want say testers to be able to push to test repos? 14:46:53 I guess we need some kind of mapping? 14:47:22 Or just protocol in requiring all groups on src.fp.o to require packager somewhere in their requirements chain 14:47:34 (and disallowing non-admins to create src.fp.o groups) 14:48:00 but we want non-packagers to be able to fork and send PR 14:48:10 well, that would work now, but at some point we are going to need to be able to allow non packagers to forks and/or test or the like 14:48:29 Right, but I don't think people will use groups for forks that often. 14:48:30 * pingou tries to think clearly 14:48:41 I'm saying just for creating groups in src.fp.o. 14:48:42 we cannot add someone to a project/package atm if they are not packager 14:48:46 there may not be a quick answer here, we could ponder it and move on for now? 14:49:13 +1 to move on 14:49:15 +1 14:49:18 so maybe we're fine, non-packager won't have direct access to the main repos 14:49:37 .ticket 6320 14:49:38 relrod: Issue #6320: is it possible to rename FAS group "cvsl10n" to "l10n"? - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/6320 14:49:56 so here the i10n group has been doing clean up... 14:50:06 however they now need to remove/add a ton of people to some groups. 14:50:23 they don't want to click buttons in the web interface, they need a small script to do it. 14:50:41 Well, they can use the standard FAS API? 14:51:17 yes, but the person filing the ticket is not a python savvy programmer... 14:51:26 they would like some help with it. 14:51:27 "easyfix" 14:51:40 i.e. tag easyfix, and get an apprentice to work on it? 14:51:52 it might not need to be python if it's a web API, right? 14:51:57 or is it a web API? 14:52:04 bowlofeggs: it's a JSON-ish API 14:52:14 (but there's simple python wrappers for it) 14:52:16 puiterwijk: we could, but it's already waited a while... ;( 14:52:18 yeah, so they can use a language they know well 14:52:48 bowlofeggs: not sure they have one :) 14:52:55 ah 14:52:57 nirik: we have apprentices in this meeting, right now. Maybe one of them wants a challenge? 14:53:08 bowlofeggs: well, other than natural languages. :) 14:53:13 haha 14:53:21 I've heard they know a lot of those over in l10n 14:53:21 they can write a script in ancient egyptian 14:53:23 sure, ok we can try easyfix... 14:53:23 Yeah agree this feels like a good easyfix ticket 14:53:37 You can assign me if you would like 14:54:02 Okay, last ticket 14:54:05 .ticket 6241 14:54:07 relrod: Issue #6241: How to orphan in EPEL only ? - fedora-infrastructure - Pagure - https://pagure.io/fedora-infrastructure/issue/6241 14:54:38 creaked: cool! please ask questions on it and thanks for taking it. 14:54:50 so, on this one... I am not sure what action is now needed. 14:54:58 document and close? 14:55:16 oh right, there was stuff. 14:55:42 so, there was suggestion about a 'fedpkg orphan' and making an 'orphan.package' 14:56:50 without per-branch permissions, what does orphaning a branch even mean? 14:57:05 it means: 14:57:22 "I don't intend to maintain it anymore, if anyone else does, please take over" 14:57:48 i.e. there's people who only care about FEdora and epel7, and will orphan el6 14:57:51 why not just e-mail the devel list? 14:57:54 I guess simply 'bugs are assigned to orphan user' 14:57:57 and if you get no response, retire it? 14:57:59 and if you orphan in EPEL it also mean: don't bug me 14:58:18 because you do care about some branches. 14:58:28 well you retire the el branch 14:58:49 it is possible to retire just the EPEL branch right? 14:59:10 yeah, I think so... dead.package in it and it gets blocked, etc 14:59:20 So: We're coming up on the top of the hour. Do we want to leave this one as a meeting ticket and pick it up next week? Or pick it up on the list? 14:59:26 you can't retire current active fedora branches tho 14:59:41 we can punt to next week. it's not urgent 14:59:49 okay 14:59:50 yeah i've got another meeting now so g2g 15:00:02 #topic quick open floor 15:00:10 we're at the top of the hour, but if anyone has anything urgent, go :) 15:00:18 nirik: we also have another meeting iirc :) 15:00:40 the pdc thing? if you and bowlofeggs are going, do I need to? 15:01:07 I'm going, but I thought I saw you on the list of invite 15:01:12 okay going to close this out. Thanks all! 15:01:13 nirik: yeah pdc 15:01:17 #endmeeting