16:00:13 #startmeeting EPEL (2014-08-29) 16:00:13 Meeting started Fri Aug 29 16:00:13 2014 UTC. The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:33 #meetingname EPEL weekly meeting 16:00:33 The meeting name has been set to 'epel_weekly_meeting' 16:00:36 * Jeff_S here 16:00:43 hi all 16:01:04 #topic meet and greets 16:01:15 nirik, can you op zodbot? 16:01:15 hello 16:01:16 I'm here, but sick, so likely less coherent than normal if thats possible. ;) 16:01:19 I'm here 16:01:31 #topic meet and greets 16:01:48 Not sure if I need to call the meeting something for the date or not 16:01:52 anyway hello all 16:01:56 o/ 16:01:59 hope you had a good week. 16:02:07 well .... it's been a week 16:02:10 :) 16:02:13 I know that some people will be joining later in the meeting and such. 16:02:37 Today's meeting is going to be 'shorter' than last weeks. I will be calling it at 17:00 UTC no matter what 16:02:57 my tweet about EPEL7 is my most successful tweet to date in terms of retweets ... so there's something https://twitter.com/TheMaxamillion/status/505362715310039040 :) 16:03:15 cool. 16:03:15 Cool. I have no idea what that means :) 16:03:19 anyhoo... sorry to go offtopic, just thought that was cool 16:03:20 * Ussat congratulates maxamillion 16:03:30 smooge: I like to think it means people care about epel 16:03:37 smooge: but I might just be overly optimistic 16:03:44 * smooge is old and can't keep his twits versus pervbooks apart 16:03:57 smooge: :) 16:03:57 or soemthing like that 16:04:13 thanks for doing that maxamillion 16:04:17 twifacespace 16:04:22 #topic Agenda 16:04:38 #info 1) Meet and Greets 16:04:45 #info 2) Agenda 16:04:54 #info 3) EPEL 7 out of beta 16:05:06 #info 4) EPEL. discussion 16:05:17 #info 5) Open Flood 16:05:24 #info 6) End meeting 16:05:37 #topic EPEL 7 out of beta 16:05:53 woot! go team! :D 16:06:01 +1 16:06:02 congrats to nirik who was very sick and dgilmore who was doing other freeze stuff on getting EPEL 7 out of beta yesterday 16:06:06 theres still a bit of bodhi work that needs to happen... 16:06:21 (right now updates aren't working right), but hopefully soon 16:06:25 oops 16:06:31 oh well it could be worse 16:06:38 indeed 16:06:38 Yep, much appreciated, thank you both 16:06:40 we didn't delete it all 16:06:41 congrats for the release on behalf of the CentOS QA team :) 16:07:20 #info Bodhi needs to get some work done still 16:07:31 #info updates do not work correctly yet but will soon. 16:07:32 * nirik will be back in a min... 16:07:50 nirik: any bodhi items that others can help? (I suspect not but figured I'd ask just in case) 16:07:53 Anything people in the field using EPEL seeing that needs to be addressed? 16:07:56 can help with* 16:08:30 smooge: nothing that's come up, I basically live in #rhel and there's a lot of EPEL users there ... not really heard of any issues thus far 16:08:38 maxamillion, I think it is that the channels are hard coded somewhere and so a patch needs to be done and then go through the alpha freeze process 16:08:45 smooge: ahhh 16:08:51 oh I forgot about #rhel 16:09:11 Thanks for putting the word there 16:09:26 smooge: I didn't ... probably should a little bit ... I'm #1 on the stats, for better or worse :/ ... http://www.delhage.se/rhelstats/ 16:09:48 but... did anyone share this news with OEL? 16:09:51 * Jeff_S hides 16:10:09 OEL knows all before we do it. It is an oracle 16:10:29 ;) 16:10:36 it's... something 16:10:49 and I guess SciLinux will need an update 16:11:05 I don't know if htey have a channel 16:11:18 I would ping tdawson but he's on PTO 16:11:21 when I updated to epel-release-7-1, it moved my previous epel.repo to epel.repo.rpmsave. will subsequent epel-release updates also overwrite the existing epel.repo files? I thought the normal procedure was to create .rpmnew files if the file was modified. 16:11:28 I'm sure they have people on the EPEL lists 16:11:48 hmmm did you edit it somewhere 16:11:55 I'll send a note, I'm on their mailing lists 16:11:58 maxamillion: the bodhi change needs to be that it looks for "EL" and it needs to look for "EL or EPEL" 16:12:02 because a .save comes up if the file was edited 16:12:15 this might be relevant if people have, say, yum-plugin-priorities installed 16:12:35 yeah, it should only do that if they were edited normally. 16:12:39 smooge: yes, I think I had edited it to include a "priority=xx" in there 16:13:12 hmm, is that file marked as %config ? If so, I'd expect it to do .rpmnew if the old file had been edited 16:13:25 one sec while I whip this out 16:13:28 For this release the URL is moving though right, so it's needed/expected 16:13:34 CentOS .repo files end up as .rpmnew files if the user's repo file has been edited in the meantime 16:13:41 but moving forward it shouldn't be replaced IMO 16:13:56 Should be %config(noreplace) then 16:13:56 avij: yeah, that's the %config macro at work 16:14:01 oh, that 16:14:29 the only actual change was 'gpgcheck=1' I think. ;) 16:15:01 note sent to ScientificLinux community http://listserv.fnal.gov/scripts/wa.exe?A2=ind1408&L=scientific-linux-users&T=0&P=14381 16:15:07 thanks max 16:15:12 smooge: certainly 16:15:19 amillion 16:15:20 I agree that for this particular case it's perfectly okay to overwrite the file to get gpgcheck=1 in there, but I'd expect that further updates to that rpm would leave user-modified repo files intact 16:16:08 avij: can you file a bug on epel-release? we can see what if anything we need to change? 16:16:28 nirik: will do 16:17:04 ok any other gotchas or problems? 16:17:22 #topic EPEL. discussion 16:17:35 the fedora ones are noreplace, so I suspect we just need to do that in epel-release too. 16:17:43 its purposly not marked as a config file 16:17:57 so when we make changes to the .repo files users get them 16:18:26 dgilmore: is it config... just not noreplace. 16:18:33 the change was to turn gpgchecking on 16:18:41 %config /etc/yum.repos.d/* 16:18:42 nirik: right thats intentional 16:18:55 we want to replace it 16:18:56 huh. fedora doesn't do it that way? 16:19:00 hrrm 16:19:04 are older ones that way? I don't like it 16:19:13 it used to and I don't remeber changing it 16:19:20 Jeff_S: yes 16:19:46 hmm, ok, let's discuss later and not hold up the meeting 16:20:01 dgilmore: thanks for clarifying 16:20:11 #action Jeff_S and dgilmore and others to discuss .rpmsave issue 16:20:27 ok back to the topic. 16:20:59 we discussed last week about various options that could be done for EPEL.<>. Was there any discussion between meetings we need to cover? 16:21:18 dgilmore, nirik ? 16:21:24 from the last meeting: https://fedoraproject.org/wiki/EPEL-faster-repo-ideas 16:21:36 there was some more posts to the lists... 16:21:45 otherwise I haven't had any more discussions 16:22:54 thank you bstinson 16:23:18 ok I saw there was a post early this morning from the stacks group. I haven't had time to respond to it yet 16:23:48 * nirik either 16:25:18 ok so first thing I want to discuess.. set up at least short term board to help guide the new repos 16:25:26 I'd say we should continue on the list (so we include the stacks folks that can'tmake it to this time) 16:25:50 need to step afk for a second, apologies 16:26:09 smooge: I thought we sort of did that last week... but I guess we never decided anything... you, me, dgilmore, and Evolution. 16:26:46 I figured it was a proposal last week. I am just cementing it this week. 16:27:36 Are there any objections to having a short term governance team containing smooge, nirik, dgilmore and Evolution? 16:28:30 no objections here 16:28:32 +1 16:29:01 nirik, dgilmore ? 16:29:11 sure. 16:29:35 #agreed EPEL short term governance team members: smooge, nirik, dgilmore, and Evolution 16:29:38 * dgilmore doesn't feel he should say yes to himself but the others yes 16:29:50 im happy to do it 16:29:54 well I was looking for a "Fuck no I won't do it" 16:30:10 anything else was implied agreement :) 16:30:17 heh 16:30:21 smooge: nice 16:31:03 Evolution agreed earlier. He is out of this meeting because it is his 4 year anniversary. 16:32:06 ok from that I hear a call to move the discussion to mailing lists to include people who could not attend this meeting due to conflicts and timezone conflicts. 16:32:51 If we do this I would like to frame what we are moving to the mailing list so I can make a report for next weeks meeting. 16:33:18 Looking at https://fedoraproject.org/wiki/EPEL-faster-repo-ideas I see several subtopics. 16:33:26 Do we want to focus on one for this week? 16:33:27 * maxamillion is back 16:34:05 good news maxamillion you were not voted off the island 16:34:54 \o/ 16:36:14 Does discussing policy, 3 repos or technical questions first makes sense to people or is there a pressing one that stands out? 16:36:29 I see Evolution got thrown under the bus, I support this +1 16:36:45 (I put that there in hopes he sees it in the chat logs ;) ) 16:37:39 i think we should work out 1.) governance structure (which we just started), and 2.) the "Policy Questions" section before we start talking about technical solutions 16:39:45 OK so my proposal is that the governance structure is a short term organization to set up policies and such in place. Its term will end 3 months after either new repos are setup or voted not going to happen by said board. If a longer term running board is required the previous board will set up policies for elections and such before ending. 16:40:20 change board to committee or whatever word works best for this org. 16:40:39 smooge: sounds great to me 16:40:46 a time limit is fine. 16:41:11 I wanted to do that so we don't end up with a lifetime FPC or people feeling they have to be here forever. 16:42:05 according to linkedin I'm still on the epel steering-whatchamicalit 16:42:10 +1 16:42:10 maybe I'll change that one day 16:42:29 Jeff_S, probably a good idea.. in case you get called on it :) 16:42:39 +1, after the policies are set it would be nice to keep some sort of group around (sort of like an outreach/quasi-steering committee) as a resource for packagers 16:42:48 EPEL Steering Committee - Dishonorable Discharge 16:43:08 EPSCO we put SCO into EPEL 16:43:34 bstinson: sure... but the same group is generally here in some unofficial capacity. But calling it out as a resource for packagers may be useful 16:43:50 Ok here were the list of policy issues from the wiki page 16:43:51 lol smooge 16:44:09 How many repos? epel, epel-rolling and epel-edge? 16:44:09 What would faster moving mean? 16:44:10 Would packages in this be able to conflict with epel packages? Base packages? 16:44:10 When would incompatible changes be allowed in each branch? 16:44:10 Different guidelines for specs/packages per branch? 16:44:10 When would a package be expired or removed? 16:44:32 any other items that people see as must haves? 16:45:02 * nirik can't think of anything else, but my head is killing me, so I'll look later when I am well. 16:45:33 nirik, ok no problem.. if you need to exit meeting.. I am planning on taking this all to the mailing list in 10 minutes. 16:46:03 smooge: i think 2-3 repos 16:46:18 and next week I will put to a formal meeting vote about the governance structure and any policy questions we discussed and wanted a vote on 16:47:34 smooge: I think thats fine 16:48:12 I figure that way people aren't pushed to decide when either ill or low on coffee 16:48:22 dgilmore, thanks 16:48:58 caffiene 16:49:01 :) 16:49:13 #info moving discussion of governance structure and policy to mailing list. Votes to finalize anything that feels need a vote will be next weeks meeting 16:49:47 #topic Open Flood. 16:50:05 OK we have ~10 minutes left. Any items for the floor? 16:50:06 ponies or unicorns for all? 16:50:14 ponicorns 16:50:19 is kbsingh around? 16:50:30 Jeff_S: not today I do not think 16:50:32 also I wanted to discuss the EPEL push-to-stable policy 16:50:42 dgilmore: he mentioned he wanted to drop by, but I guess not 16:50:50 ok Jeff_S 16:50:55 you have the conch 16:50:56 Jeff_S: what would you like to see changed? 16:51:35 It'd be nice to have a way around the two week in testing policy without explicit approval -- even if it's maybe for a group of "proven packagers" or whatever 16:52:03 Jeff_S: if you get enough karma it goes stable 16:52:08 Do we have data to show that stuff actually gets tested while in "testing"? I expect it doesn't happen much 16:52:18 not sure 16:52:19 dgilmore: so for a new package nobody cares about? Or some other leaf package 16:52:33 two weeks sitting there with no feedback is pretty useless IMO 16:52:56 I understand wanting to keep EPEL stable, but it'd be nice to have a way around it in some cases 16:53:36 Jeff_S: sure, maybe we can work with the qa guys to get some taskotron tasks running and allow it to go after passing some tests 16:53:40 well I was thinking we would look at it with the epel-rolling to help get more views 16:54:29 maybe we put them into a faster moving repo 16:54:37 then move to epel after some time period 16:54:41 there have been 24 karmas for el6 packages this month. pretty sad. ;) 16:54:48 heh, thanks for looking nirik 16:55:14 previous 2 were over 50 each 16:55:22 I'd propose that one week in testing is sufficient -- isn't that the current Fedora policy? 16:55:29 guess people who give karma moved to 7 16:55:34 hmm, maybe 16:55:47 that's actually more than I expected. But for how many package updates was that? 16:56:14 Jeff_S: hi 16:56:27 kbsingh: just wanted to ping you since I know you had something you wanted to discuss 16:56:50 I've had a series of meetings that overran 16:57:04 so in july there was 54 karmas for el6 packages, on 35 packages. 16:57:33 Jeff_S, I will add that to policy questions on list 16:57:39 smooge: thanks 16:57:59 #action Add time in testing to policy discussion on mailing list 16:58:05 there were two things that I wanted to bring up actually, is now a good time ? 16:58:25 kbsingh: yep 16:58:37 it is open floor. I was hoping to end the meeting in 2 but will delay 16:58:43 so, 1) was how do we handle overlapping content between EPEL and CentOS ( SIG's mostly ) 16:59:04 eg. cloud-init, we've got opennebula and cloudstack support rolled into the 0.7.5 version 16:59:07 undecided, but we should look at it 16:59:31 iirc, the last epel7 version was 0.7.4 16:59:33 kbsingh: i dont think epel7 had cloud-init as its part of rhel 16:59:41 i could be wrong 17:00:05 dgilmore: its been a painful conversation - *I* think it is a part of rhel, but rhel folks seem to think they just use it for some of their stuff, and its not really RHEL 17:00:19 cloud-init-0.7.5-6.el7 17:00:24 the idea of RH-COMMON for 7 was floated... but is deadending so far 17:00:30 there is a 0.7.5 build in epel 7 17:00:42 ah interesting 17:01:18 although that is going to cause some level of grief for people who use centos cloud images, unless we upstream the local patchs from here as well 17:01:32 so thats an example... I am sure this is going to be more of an issue ongoing 17:01:39 and we should have a plan / process to handle it 17:01:42 so in the past, I would have said that EPEL doesn't conflict with CENTOS-Core (if its not in Extras, Plus, etc we don't conflict) 17:01:43 it needs some coordinating on some level 17:01:58 either remoaval from epel or package syncronisation or something 17:02:20 smooge: in most cases, this isnt core content ( eg. cloud-init is CloudInstance SIG ) 17:02:39 removal might cause problems for other epel packages, but yeah... 17:02:39 dgilmore: initially, if we can find a way to sync, that would be better 17:02:43 * nirik nods 17:02:48 kbsingh: I am okay with that? 17:02:55 the challenge is going to be patch sync. 17:03:18 so this also leads up nicely to point 2) that i wanted to bring up 17:03:24 yeah.. so in the new now.. I would like to see a feed system where sigs work on feeding stuff into EPEL-rolling/whipcrack or whatever its called 17:03:28 if we have the same people maintaining both, they could just keep them in sync? 17:03:36 while i dont know for sure myself, mikem23 noted that its possible to send karma via an api call 17:03:55 yeah bodhi does have an api 17:04:04 and there is a tool called fedora-easy-karma 17:04:14 so what i want to propose is that we start running ci jobs for epel as well, and start sending some karma down from the ci runs 17:04:20 which will provide karma for builds installed in testing 17:04:27 initially it would be only rpm level, but we can easily expand that to include some functional stuff as well 17:04:54 maybe we can start with simple things like 'new package upgrades previous' and 'md5's for common content in 32bit/64bit match, etc 17:05:03 I like this :) 17:05:07 kbsingh: sure :) 17:05:19 cool. whats the ci system? 17:05:45 nirik: ci.dev.centos.org is where we run a few things, there is a bigger, better system under consideration.. 17:05:57 nirik: or, if your question was 'what' then - Jenkins 17:06:05 ok. 17:06:30 what we will need is someway to get a ping back or a call back with a url that has 2 things in it : name of package, url to just this build. 17:07:07 or well, something - we dont need to solve mechanics right now. but if the idea is acceptable, then we can start working out the mechanics on a list 17:07:44 Yeah, we're running long, let's discuss on list 17:07:44 kbsingh: sure, more automated testing is a win 17:07:50 I think it's a great idea 17:08:03 yeah. we do have fedmsg possibly for that info 17:08:12 right, and i think the epel group is small enough that we can actually agree to something 17:08:22 hahahaahaa 17:08:28 oh sorry was that out loud 17:08:31 you can at least get the koji task from fedmsg, then you can get teh rest from koji's api 17:08:36 lol 17:09:27 do we want to discuss some of this on centos-devel ? or epel-devel ? we have quite a few people on centos-devel and if you guys dont mind, thrashing it out there might be nice 17:09:48 reminds me i need to sign up for centos-devel and raise some issues 17:10:03 dgilmore: there are no issues, only features :) 17:10:08 some more enjoyable than others 17:10:27 kbsingh: well some features are undesirable ;) 17:11:00 * nirik is on both lists 17:11:04 some of it is working out better CentOS EPEL interactions and simplifing contribution 17:12:19 I've gotta run... will follow on list(s) 17:12:26 okay 17:12:37 same here. its almost 6:30pm, I'm about to be attached by a 2yr old any second now 17:12:42 I think we are about at the point of wrapping up 17:12:45 thanks guys 17:12:49 fyi, the aforementioned epel-release bug (if it is a bug, tbd) is at https://bugzilla.redhat.com/show_bug.cgi?id=1135576 17:13:04 I'll start some conversations with context and examples on those two points 17:13:23 ok anything else? 17:13:40 stopping in 30 seconds 17:13:57 #topic Thanks for coming everyone. See you on the list. 17:14:12 #endmeeting