16:04:26 #startmeeting RELENG (2019-07-04) 16:04:26 Meeting started Wed Jul 3 16:04:26 2019 UTC. 16:04:26 This meeting is logged and archived in a public location. 16:04:26 The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:04:26 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:04:26 The meeting name has been set to 'releng_(2019-07-04)' 16:04:26 #meetingname releng 16:04:26 The meeting name has been set to 'releng' 16:04:26 #chair nirik sharkcz pbrobinson puiterwijk mboddu dustymabe ksinny jednorozec 16:04:26 Current chairs: dustymabe jednorozec ksinny mboddu nirik pbrobinson puiterwijk sharkcz 16:04:26 #topic init process 16:04:39 morning 16:04:44 nirik: Hello again 16:04:46 :) 16:05:20 :) 16:05:24 * mboddu haven't looked at the tickets, lets see 16:06:40 Lets start with epel8 update 16:07:18 #info fedpkg and fedscm-admin changes are tested in stage and are working. People can request epel8 branches just like they do for fedora release branches. 16:07:26 #undo 16:07:26 Removing item from minutes: INFO by mboddu at 16:07:18 : fedpkg and fedscm-admin changes are tested in stage and are working. People can request epel8 branches just like they do for fedora release branches. 16:07:32 #topic EPEL8 16:07:34 #info fedpkg and fedscm-admin changes are tested in stage and are working. People can request epel8 branches just like they do for fedora release branches. 16:08:05 #action mboddu need to create pungi configs and scripts to run epel8 nightly composes 16:08:32 nirik: ^ Where do we keep this config? pungi-fedora repo and epel8 branch? 16:08:40 well, we aren't ready for branches in prod yet right? 16:09:00 yep. I would say so... or master branch in a epel8 config... 16:09:39 nirik: On dist-git prod, yes, we are not ready, I tested them on stg only 16:10:07 And I prefer the epel8 branch to just keep the things separate as we do with branched composes from rawhide composes 16:11:24 ok, that works 16:12:15 nirik: Couple of questions on epel8 16:12:30 Did we ever provide upgrade path from epeln-1 to epeln? 16:12:58 well, not sure what you mean there... 16:13:20 so say we have epel8 and epel8-playground or whatever we call it (rawhide is bad we should pick another name soon) 16:13:28 and 8.1 comes out. 16:14:05 we then take our existing packages, tag them all epel8.0 and move them to a epel8.0 repo, never to be updated again. 16:14:27 then our first compose with 8.1 populates everything in epel8 again... 16:14:33 and updates that, etc. 16:15:06 nirik: Yes, I got that, what I asked is from epel7 to epel8? 16:15:25 ah... no. 16:15:44 we have never cared about upgrade path between major versions. 16:15:59 Okay, good to know :) 16:20:54 nirik: In your above plan, we dont need epel8-playground then, right? Since we clone epe8.0 from epel8 when 8.1 comes out and 8.1 points to epel8, right? 16:21:48 well, not for that, but as a place to experement/rapidly push changes, etc it's useful 16:22:31 so you push your changes to there first, test, iterate, then if you think/want it to just go to stable mode, you build for epel8 branch that exact thing and it goes out via bodhi, etc. 16:22:49 but it's all to be figured out 16:23:41 So, what does epel8-playground buildroot comes from? When 16:23:48 1. 8.0 is the only release 16:23:58 2. 8.1 is released 16:24:14 epel8-playground and epel8 buildroots, both? 16:24:24 I would think it would have it's own buildroot that is like rawhide... so after you build it updates. 16:25:05 nirik: Then why cant the maintainers use -override tag to achieve the same thing? 16:25:18 If it works, thats great, if not, remove the override and build the stable one 16:27:59 well, the playground rpms would be published/shipped out to users to use... you don't want things building against them until you are ready... 16:28:54 so say... 8.0 has Xfce 4.12 in it... (the current stable release) 16:29:18 nirik: Okay, so it will separate repos coming out for epel8 and epel8-playground, even though they might be using the same buildroot 16:29:19 ? 16:29:20 4.14 rc's are coming out. You would want to build them in playground, and get feedback testing/ 16:29:49 but thats a stack of like 20 packages you would need overrides for, and if you left one and accidentally had to push a bugfix of 4.12 out it could get built with the wrong thing 16:30:13 yeah, well, I am not sure they can share a buildroot, but perhaps it could work. 16:30:23 but yes, otherwise seperate... 16:30:56 except due to the fedpkg config it will normally be the same git hash/tree unless someone specificallywants to make a playground build. 16:31:38 nirik: Yes, got it, thanks for explaining it to me, I was a bit confused whats the need for epel8-playground 16:32:37 same buildroot = coming from rhel8 is the same, but epel8 might have bodhi overrides and epel8-playground will have playground builds 16:32:41 same as rawhide, a place to do integration work/play with new versions/ship things that are not stable at all/ship things you don't want to support very long 16:33:17 yeah, but we don't want to build stable stuff from the playground builds... so I think it would be better for them to be seperate... 16:33:24 Thanks nirik for the info 16:33:26 I mean they inherit the same rhel8 base... 16:33:36 but then epel8 just has stable updates + overrides 16:33:48 and epel8-playground has all builds 16:35:21 Right, thats what I meant 16:36:07 Okay, moving on 16:36:16 #topic #7400 Rethink how we handle networking config with createImage (oz / ImageFactory) 16:36:21 #link https://pagure.io/releng/issue/7400 16:36:37 nirik: Any idea if it has been fixed? 16:36:48 * nirik looks 16:37:22 no, I don't think thats done. 16:37:28 we need the actual koji patch. ;) 16:38:07 "We could send a PR for Koji to use this to disable 'predictable' interface naming during the image creation; either to just always include a kernelparam value of biosdevname=0 net.ifnames=0 for createImage tasks, or to make this configurable via the command line somehow and then have pungi / pungi-fedora pass it in for the images we want to use it for (handwave handwave)." 16:39:15 so, I guess we should file a koji bug asking... and/or find someone to write the patch 16:40:37 nirik: Okay, I will file a bug against koji, thanks 16:40:47 cool. 16:42:46 #topic Alternate Architecture Updates 16:43:15 nirik: I heard there was some moment in s390x issue while you guys are in F2F meetings, any update on that? 16:44:42 so, we have a s390x buildvmhost now, and I have configured it. I have been trying to setup a guest builder on it, but networking is not working right. ;( 16:44:55 once networking is sorted, I hope to move all the existing builders to it. 16:47:05 #info Now we have a s390x buildvmhost now, and nirik have configured it but it has networking issues, once its sorted out, nirik will try to setup a guest builder on the buildvmhost and once tested, all the existing builders will moved 16:47:17 #topic Open Floor 16:47:22 Anything? 16:47:24 its quite noticabily faster... 16:47:38 * mboddu looking forward for it 16:48:47 me too 16:49:48 Well, if nothing, lets close the shop and give back the 10 min 16:49:52 Thanks nirik for joining 16:50:44 #endmeeting