15:33:48 #startmeeting RELENG (2016-03-21) 15:33:48 Meeting started Mon Mar 21 15:33:48 2016 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:33:48 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:33:48 The meeting name has been set to 'releng_(2016-03-21)' 15:33:48 #meetingname releng 15:33:48 #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion 15:33:48 The meeting name has been set to 'releng' 15:33:48 Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll 15:33:51 #topic init process 15:33:56 who is her 15:33:58 here 15:34:01 morning 15:34:04 * sharkcz is here 15:34:06 .hello maxamillion 15:34:07 maxamillion: maxamillion 'Adam Miller' 15:34:08 * nirik runs to get a cup of coffee 15:36:17 * pbrobinson is sort of here 15:38:24 lets get started 15:38:37 there is no tickets with meeting keyword 15:38:48 we really need to go through the open tickets soon 15:38:59 #topic Secondary Architectures updates 15:38:59 #topic Secondary Architectures update - ppc 15:39:05 pbrobinson: how is ppc? 15:39:16 fine 15:39:23 rebuild is done 15:39:29 pungi4 is running nightly OK 15:39:37 got a few fixes to do this week 15:39:44 * masta is here 15:39:46 and to add some other bits 15:39:57 other than that it's looking reasonably good 15:41:09 cool 15:41:26 and other than the hub infra is all done? 15:41:32 yep 15:41:47 so hub will be rebuilt post alpha 15:41:51 awesome 15:41:59 and I'll be reinstalling all the power8 boxes too 15:42:11 have we started to look at teh shared koji-shadow box yet? 15:42:15 but that's all just ansible so that's just standand and no outage 15:42:40 I have a sort of shared shadow box running but not in Fedora infra yet. sort of a prototype 15:42:53 post alpha I'll be dealing with that properly as well 15:43:14 awesome 15:43:25 #info ppc is moving along well 15:43:49 #info ppc infra is all in ansible except for the hub (It will be done post alpha) 15:44:04 #topic Secondary Architectures update - s390 15:44:10 sharkcz: how is s390? 15:44:42 so it's OK 15:44:44 except pungi4 we are looking good, largely up-to-date with f24/f24-updates-testing, f25 builds have also started 15:45:03 I want to move over to a pungi4 Everything only compose soon 15:45:18 but I'll work with sharkcz to deal with that soon 15:45:29 ok, to have the repos soon 15:45:37 yep 15:46:11 as a short term thing we should be able to turn off building installable media 15:46:21 but we will need to find an answer for that 15:46:42 yes, going to work with sharkcz post alpha 15:46:48 ack 15:47:35 so pungi4 for s390 will be a two phase approach with 1 RSN and 2 a little later 15:48:15 right, for beta we want to have it somehow solved 15:48:26 yep 15:48:47 #info s390 builds going well 15:48:55 I have some ideas on that, post alpha I'll have some breathing room to test my theoryies 15:49:01 #info we need to find an answer for building install media 15:49:42 sharkcz: do we have qemu or hercules docs somewhere? 15:49:57 and does one of them work better than another? 15:50:10 if I wanted to setup something in my test environment at home? 15:50:20 dgilmore: dunno how well can qemu do s390 umaltion, but for hercules we have something on my blog 15:50:41 s390 emulation 15:51:29 http://sharkcz.livejournal.com/12268.html 15:52:20 sharkcz: okay, I thought I read somewhere that qemu-system-s390x was a better way to go 15:52:53 anyway we can take this out of the meeting 15:53:10 dgilmore: yep 15:53:15 dgilmore: I know qemu does the job well for kvm, but no idea how it does full system emulation 15:53:30 yep, just ping me if you have questions 15:53:36 sharkcz: will do 15:53:38 #topic Secondary Architectures update - arm 15:53:43 pbrobinson: how is aarch64? 15:53:50 good 15:53:56 similar status to ppc 15:54:06 we have pungi 4 nightly composes 15:54:11 #info aarch64 is similiar to ppc 15:54:17 #info pungi4 composes 15:54:32 pbrobinson: have we tried to do an alpha compose for either? 15:54:33 looking to hack in a lmc add a nightly minimal image for it this week 15:54:44 dgilmore: tomorrow likely 15:54:51 cool 15:54:56 I was waiting to see what the state of 1.6 15:55:02 its not good 15:55:15 I figured 15:55:30 was going to do alpha status next 15:55:41 I took a few days off pungi4 once I got the nightlies working 15:56:08 this week is back to it to clean up a bunch of things on nightlies and look at adding things like qemu/docker/images etc 15:56:15 across both ppc/arm 15:56:25 and get alpha releases in order 15:56:25 should not be hard 15:56:36 no, it's not, it just needs testing and time 15:56:49 * nirik notes the ppc/arm pungi4 crons are going to root@ still. ;) 15:57:34 nirik: yes, I know, it's on my list for this week 15:57:48 no worries. 15:57:56 nirik: it's erroring out on line 25 for some reason, that line is nothing 15:58:01 I've got a few bugs to work out 15:58:16 let me know if I can assist any 15:58:23 I needed some time out from it all, it was sending me a little more mad that usual last week ;-) 16:01:32 anything else on arm? 16:02:02 nope 16:02:12 #topic Alpha Status 16:02:28 so right now we are expecting another compose request 16:02:44 the background bug was not fixed 100% 16:03:00 dgilmore: background bug? 16:03:05 a couple of places needed updating to pull in the new backgrounds 16:03:48 https://bugzilla.redhat.com/show_bug.cgi?id=1319043 16:04:22 dgilmore: +1 thanks 16:04:36 there is a few proposed blockers also 16:04:44 https://qa.fedoraproject.org/blockerbugs/milestone/24/alpha/buglist 16:04:59 Blocker bug meeting going on right now in #fedora-blocker-review 16:05:13 sharkcz: we know 16:05:17 sgallagh: we know 16:05:20 shorry sharkcz 16:06:03 One thing I really do not like about the new compose process 16:06:17 I have to edit the config every time we do a compose 16:06:29 dgilmore: really? 16:06:31 dgilmore: why so? 16:06:33 I will have to file some RFE's so we can not do that 16:06:41 +1 16:06:45 that does not sound ideal 16:07:14 maxamillion: see https://pagure.io/pungi-fedora/7e9270e19193d6980b6f8c9968b1c29e51378062 as an example 16:07:28 dgilmore: ah ... ouch 16:08:06 dgilmore: I mean ... it kind of makes sense, similar to how packagers have to bump the Release ... but :/ 16:08:12 other than some weird issues with koji tasks though it has been pretty smooth 16:08:39 that's good 16:08:40 maxamillion: yeah I think a CLI flag to set it would be better 16:09:05 +1 16:09:27 potentially we would also want to set the git hash for spin-kickstarts 16:09:41 +1 16:09:50 yeah, I could see that being useful 16:10:41 still not had much feedback 16:10:53 oh one other thing 16:11:10 due to kernel bugs on i386 all 32 bit x86 images are turned off 16:11:33 we are not making lives or cloud images 16:11:56 I am going to work this week on some openQA integration 16:12:18 so we only push composes live after they have passed some amount of automated install testing 16:12:36 that will include rawhide and possibly branched 16:13:46 #info no 32 bit x86 images 16:13:59 #info alpha is going reasonably well 16:14:16 dgilmore: the openqa integration would be great 16:14:21 #info looking at gating composes based on openQA results 16:14:25 rawhide? that would be adding gating to rawhide composes? 16:14:35 nirik: yeah 16:14:51 cool. I'm on board with that. We could also make sure everything is signed. 16:15:01 (well, after we have a f25 key and sign everything) 16:15:01 right now if rawhide is not installable it fails the compose 16:15:14 nirik: yeah I need to do that today 16:15:44 #action dgilmore to make f25 keys 16:16:04 dgilmore: we have a hardware box for autosign02... I've done nothign at all with it yet. 16:16:09 nirik: so I figure if the compose has to be installable, lets make sure it works 16:16:29 nirik: yeah we should sit down and figure the pieces out for that 16:16:41 yep. so much to do, so little time. 16:16:47 since we will need some dev work to enable signing requests 16:17:08 so we can sign repo metadata etc 16:17:31 * nirik nods. 16:17:32 anyway that is it from me for alpha 16:18:07 #topic open floor 16:18:15 * sharkcz has 3 items 16:18:16 does anyone want to bring something up? 16:18:22 sharkcz: the floor is yours 16:18:22 lmacken: you make any progress with faster composes for bodhi? 16:19:25 1st - the empty buildroot info - I was updating s390 builders, saw the same issue, but at the end it seems that a missing restart o fkojid was the culprit, now the builders work fine with latest f23 rpm and koji from the infra repo 16:19:44 huh. weird. 16:19:44 would be great if someone could confirm that on the staging koji 16:20:05 sharkcz: weird 16:20:10 I agree 16:20:47 sharkcz: I need to do some work on stg koji today 16:20:52 I will do some testing 16:21:38 dgilmore: thx, I havent' checked the koji spec, but if I did explicit "systemdctl restart kojid", it started to work correctly 16:22:26 2nd - for the weak deps - I'm working with the dnf guy, so hopefully it will start moving ahead too (bug 1300669) 16:22:46 sharkcz: koji used to restart on upgrade, but that stoped happening with teh move to systemd 16:23:08 sharkcz: I need to follow up with tibbs|w about what the policy should be 16:23:38 #action dgilmore follow up with tibbs on the fpc ticket for how weak deps should be handled in the buildroot 16:23:39 dgilmore: ok, and what exceptions are allowed 16:23:42 dgilmore: I think everything FPC had to add was in that ticket. 16:23:51 ah, it was for weak deps 16:24:19 tibbs|w: which ticket? 16:24:27 I went looking last week and couold not find it 16:24:49 Didn't you file one for the FPC? I went on vacation and now I can't remember the state of things. 16:25:00 dgilmore: my goal is to confirm with the dnf guys that all arches behave correctly and the same 16:25:03 tibbs|w: I do not know 16:25:13 OK, I'll try to swap all of that back in. 16:25:51 their functional testing was limited to x86 16:25:53 I do remember we talked about it and basically the thing is that if you really need something in order to build, you had damn well not depend on anything following weak deps. 16:26:07 tibbs|w: maybe I forgot to file a ticket 16:26:09 That's really the only thing that makes sense. 16:26:25 tibbs|w: that is fine. 16:26:28 I know there was some FPC discussion at least. I will try to dig it up. 16:26:37 thanks 16:26:55 sharkcz: :( 16:27:05 sharkcz: we will figure out what we should be doing 16:27:20 but I suspect that it will be making sure weakdeps is turned off 16:27:47 sharkcz: what was the third thing? 16:28:29 last is integrating srpm-excluded-arch.py back to the compose/mash process so we have up-to-date list for koji-shadow 16:28:42 sharkcz: it is, its just not quite right 16:28:47 it is producing empty files 16:28:54 ah 16:29:32 https://kojipkgs.fedoraproject.org/compose/branched/Fedora-24-20160321.n.0/logs/ 16:29:43 the excludearch files there 16:30:00 critpath needs fixing also 16:30:03 I see 16:30:32 * linux-modder super late sorry 16:31:01 #info there is some fixes that need to be made in compose scripts, patches welcome 16:31:07 linuxmodder: almost done :) 16:31:12 but welcome 16:32:10 does anyone have anything else for open floor? 16:33:16 okay lets wrap up 16:33:20 #endmeeting