15:34:40 #startmeeting 15:34:40 Meeting started Mon Apr 28 15:34:40 2014 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:34:40 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:34:47 #meetingname releng 15:34:47 The meeting name has been set to 'releng' 15:35:19 #chair dwa nirik sharkcz tyll 15:35:19 Current chairs: dgilmore dwa nirik sharkcz tyll 15:35:43 #topic roll call 15:35:47 who is here? 15:35:54 * tyll is 15:37:10 #chair masta 15:37:10 Current chairs: dgilmore dwa masta nirik sharkcz tyll 15:37:12 * sharkcz is here 15:38:15 morning 15:38:36 lets get started 15:38:54 #topic tickets https://fedorahosted.org/rel-eng/report/10 15:40:14 one sec need to haul out some rubbish 15:44:43 sorry done 15:45:00 #topic https://fedorahosted.org/rel-eng/ticket/5854 15:45:03 masta: ? 15:47:50 #topic https://fedorahosted.org/rel-eng/ticket/5362 15:48:13 #info updating pungi and mash is on my todo list for this week 15:48:28 #topic https://fedorahosted.org/rel-eng/ticket/5846 15:48:49 cool. 15:48:58 bochecha is not here 15:49:19 I think the last plan I saw was fine... I'd be good with moving ahead with implementing it. 15:49:24 or did we have outstanding questions? 15:50:11 I noticed there was one mentioning of MD5 15:50:52 https://github.com/tyll/fedora-infrastructure/commit/5af9e96d670197ba8ec968bfbea32057864542fd 15:51:01 +unneccessary 15:52:06 A timeline is missing, e.g. when do we patch which part 15:52:27 yeah we dont have an implementation timeline 15:53:03 Is there some kind of staging system to test the patches? 15:53:30 nirik: we have a staging pkgs right? 15:53:41 yep. 15:53:48 we need to sync it 15:53:56 but it's there. 15:53:56 okay 15:54:05 So the next step would be to test it there imho 15:54:11 * nirik nods 15:54:16 so yeah we should get stg done 15:54:43 pkgs.stg should be pretty complete. 15:54:56 it's the koji / builder part thats not really there 15:55:07 #info test patches in staging pkgs 15:56:08 #action get stage converted 15:56:29 #info test everything then work out timeline to migrate production 15:57:35 anyone have anything else here? 15:57:47 I don't 15:58:41 * nirik doesn't 15:58:42 anyone want to commit to converting stage? 15:59:38 okay lets move on 15:59:42 #topic https://fedorahosted.org/rel-eng/ticket/5870 15:59:47 nirik: any updates? 16:00:10 not really, but would be nice to try and move it forward. ;) 16:00:18 did we decide what key(s) we would use? 16:00:31 could we then look at doing a 'best effort' manual signing as a stopgap? 16:00:37 well sigul cant do it right 16:01:07 we could do manual signing with the f21 key today 16:01:16 but packages will get out unsigned 16:01:24 we can setup a fedmsg listening thing to sign with... but it would take some work to sigul (which mitr can't commit to, but perhaps we could find someone else) 16:01:27 right 16:01:34 just a best effort type thing. 16:02:45 * nirik updates the ticket, since I failed to 16:02:58 http://meetbot.fedoraproject.org/teams/releng/releng.2014-03-31-15.30.html 16:03:19 at branch, f21 gets all f21 signed and rawhide gets all f21 signed from there, but f22 after that (tyll_, 15:55:24) 16:03:57 * nirik nods. 16:04:15 so, we could sign them all with f21 now, but there will always be some not yet signed. 16:04:34 we could gate them, but that could be anoying to people... 16:04:55 ie, builds go into f21-pending, and when we sign them, we move them to f21 16:05:47 would add a possible delay to some builds getting out 16:05:51 I can update the ticket with those ideas and we can ponder on it? 16:05:53 dgilmore: yep. 16:06:13 some people think things are too slow already 16:07:33 not sure whether gnome-software supports it, but an iterim step would be to sign the repodata, so packages can be verified against the metadata 16:07:35 yeah, impatience is common. ) 16:07:50 then only one additional signature would be required 16:07:58 tyll: has the same disadvantage as gating... need someone to sign every compose repodata 16:08:27 tyll: signing the metadata adds manual intervention at the end of the process 16:08:43 and would require us to redo all of our processes 16:08:52 ok 16:09:48 unless I'm missing something (possible) we already do get the metadata via https to mirrorlists (metalink) 16:09:56 i honestly don't care if gnome-software supports it or not. i think it would depend on the backend 16:10:01 yum supports it 16:10:05 not sure about dnf 16:10:36 I don't think it gets us much... signing packages is better for being able to verify them after the fact, etc 16:12:37 right 16:13:21 anything else here? 16:14:08 lets get some feedback if the manual stuff would be worth doing... 16:14:11 I updated the ticket.. 16:14:39 okay 16:14:47 #topic secondary arches 16:14:53 #topic secondary arches - s390 16:15:01 sharkcz: got an update for us? 16:15:33 not much, working on partial rebootstrap due glibc ABI breakage, starting with binary perl modules 16:15:43 fun 16:16:14 yep, it took quite long to realize what went broken, what are the consequences 16:16:48 but the f21 mass rebuild will clean up everything 16:17:01 so it's not that bad :-) 16:17:50 cool 16:17:58 #topic secondary arches - ppc 16:18:08 dwa: got an update for us? 16:18:59 I think there are just the regular breakages big vs. little endian, etc, webkit and others 16:19:17 okay 16:19:30 #topic secondary arches - aarch64 16:19:48 aarch64 is moving along pretty well 16:19:59 its catching up to primary 16:20:17 most things are bootstrapped 16:20:37 gnat i think will be bootstrapped soon 16:21:05 #topic openfloor 16:21:22 anyone have anything to bring up? 16:21:53 just a FYI about storage... 16:21:56 https://bugzilla.redhat.com/show_bug.cgi?id=1091708 - createrepo in rhel7 is broken for koji with external repos, I've cloend the fedora bug to el7 16:22:15 sharkcz: okay 16:22:23 nirik: any progress? 16:22:26 we are going to be migrating our storage tomorrow night. There's not supposed to be any outage, but we will see... after that hopefully things like rawhide composes should be much faster 16:23:02 that woudl be nice 16:23:53 we will see. ;) 16:24:15 indeed 16:24:23 anything else? 16:24:34 if not will end the meeting 16:26:15 ah geez.... sry I'm late 16:27:12 #endmeeting