14:00:18 <bstinson> #startmeeting CBS/Infra 14:00:18 <centbot> Meeting started Mon Mar 2 14:00:18 2015 UTC. The chair is bstinson. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:18 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic. 14:00:25 <bstinson> #meetingname CBS/Infra 14:00:25 <centbot> The meeting name has been set to 'cbs/infra' 14:00:44 <bstinson> #chair alphacc Arrfab kbsingh MerlinTHP 14:00:44 <centbot> Current chairs: Arrfab MerlinTHP alphacc bstinson kbsingh 14:00:55 <kbsingh> hello 14:01:02 <Arrfab> hey 14:01:18 <MerlinTHP> heya o/ 14:01:38 <bstinson> #topic Status Updates 14:02:01 <bstinson> #info centpkg: not much progress 14:02:18 <bstinson> i need to get with kbsingh soon to button up some of the lookaside stuff 14:02:27 <hhorak> Hey 14:02:52 <hhorak> bstinson: lookaside is quite important for SCLo SIG :) 14:03:11 <bstinson> indeed :) 14:03:54 <kbsingh> sounds good 14:05:04 <bstinson> anything going on with central auth? 14:05:33 <MerlinTHP> ping Evolution re: above 14:06:11 <kbsingh> i was away last week so still catching up. but havent seen anything fly past around auth as yet 14:06:14 <Evolution> not currently. I'm mostly beating on aarch64 and systemd 14:07:13 <kbsingh> do we have 1 person who owns the auth story? 14:07:46 * MerlinTHP passes Evolution a stick 14:07:53 <Arrfab> kbsingh: when we tried to discuss that publicly on the -devel list, we had no real reactions/traction it seems 14:08:36 * Evolution nominates bstinson 14:08:52 * bstinson can take that up 14:08:58 <kbsingh> ok 14:09:21 <bstinson> Evolution: you free this week for a brain dump? 14:09:26 <Arrfab> the problem is that if someone needs to write code, he'll have to maintain that code for the lifetime, and also do some serious security audit on it 14:09:28 <Evolution> yep. 14:10:22 <kbsingh> bstinson: put it up at wiki.c.o then so we can socialise things around 14:10:49 <bstinson> i think any solution we pick will require some code, but we should pick the solution that requires the least 14:11:03 <kbsingh> Arrfab: sure but the feature owner will typically not be the one writing the code ( unless they want to ) 14:11:05 <mikem> good morning 14:11:12 <mikem> (school dropoff ran long) 14:11:20 <kbsingh> mikem: moin 14:11:40 <Arrfab> #link http://wiki.centos.org/InfraWiki/CentralizedAuth 14:12:19 <Arrfab> (as a reminder) 14:12:50 <bstinson> thanks, we'll add stuff there 14:12:58 <kbsingh> Arrfab: that would be a good place to put things into 14:13:21 <kbsingh> lets start by writing down the story and req 14:18:06 <bstinson> Arrfab: can you give the appropriate edit permissions on that page? or should that request go to -docs? 14:18:59 <Arrfab> #action Arrfab to give Brian edit rights on the auth wiki page 14:19:27 <bstinson> awesome 14:19:56 <bstinson> other updates? anything going on in koji-land? 14:20:28 <kbsingh> are we at open floor stage ? 14:20:37 <Arrfab> bstinson: alphacc and myself have updated koji last week on all cbs nodes, and now it seems imagefactory can build vagrant boxes 14:21:11 <bstinson> #info koji: Imagefactory can build vagrant boxes now! 14:21:15 <bstinson> woohoo! 14:21:38 <kbsingh> is that just for atomic or all kinds ? 14:22:08 <Arrfab> kbsingh: afaik just atomic right now 14:22:44 <bstinson> kbsingh: i was going to see if we can schedule a workflow hangout, and then do open floor 14:23:08 <kbsingh> ok 14:23:41 <kbsingh> Arrfab: and that consumes the nightly atomi ostree repo and lorax runsfrom teimzul nightly? 14:23:53 <kbsingh> these would be the buillogs bits.. 14:25:17 <kbsingh> erm... 14:25:37 <kbsingh> Arrfab: are those vagrant box's built from the buldlogs content generated by reimzul nightly ? 14:25:39 <Arrfab> kbsingh: http://cbs.centos.org/koji/taskinfo?taskID=7591 14:26:28 <Arrfab> kbsingh: and ostreesetup --osname="centos" --remote="installmedia" --ref="centos/7/atomic/x86_64/cloud-docker-host" --url="http://buildlogs.centos.org/centos/7/atomic/x86_64/repo/" --nogpg 14:26:35 <Arrfab> so I'd say "yes" 14:27:13 <kbsingh> but the builders cant see the internet right ? so this wold be / shold be the mirror'd content locally 14:27:47 <Arrfab> kbsingh: yes 14:27:59 <kbsingh> so if this is working, can we then say that - the nightly reimzul runs work, the sync to buildlogs works, the sync down tothe builders works, then the koji instance can consume it 14:30:19 <Arrfab> bstinson: something to discuss for the cbs meeting ? 14:31:54 <bstinson> Arrfab: at the next one you mean? 14:32:48 <alphacc> nothing to report on my side 14:33:00 <bstinson> kbsingh: we can try to grab imcleod who can verify if the bits on his end worked 14:33:07 <kbsingh> ok 14:33:42 <bstinson> would be neat to get a short walkthrough of that feature next time 14:34:25 <bstinson> if there are no other status updates, 14:34:56 <bstinson> #topic Scheduling a SIG hangout to talk about workflow(s) and what else needs implemented 14:36:03 <bstinson> should we schedule this during one of our CBS meeting times? 14:36:09 <kbsingh> sure 14:37:17 <bstinson> I'm thinking either the 16th or the 23rd 14:37:56 <kbsingh> 16th might be easier for me, i might be on the road on the 23rd ( but unconformed ) 14:38:05 <kbsingh> unconfirmed 14:39:13 <bstinson> that should work, we should get the word out on -devel and to the sigs. it would be nice if each one had someone there 14:39:27 <kbsingh> for the 16th ? 14:39:36 <bstinson> yeah 14:39:42 <kbsingh> cool 14:41:10 * bstinson is calculating for daylight savings time 14:42:19 <bstinson> #topic Open Floor 14:43:08 <kbsingh> 2 things from my side : 14:43:28 <kbsingh> 1) there is movement on armv7 in the arm-dev@centos.org list, MerlinTHP is mostly point for it ( increasingly ) 14:43:44 <kbsingh> we should reach out and see what they are doing, how they are doing it, and what the cbs role might be ( even if its not right away ) 14:43:46 <MerlinTHP> Nearly there with the initial build. 14:43:59 <MerlinTHP> Just two packags from the c7 GA set not building. 14:44:08 <MerlinTHP> Both test suite failures. 14:44:20 <kbsingh> secondly, powerpc for power6 / power7 is mostly done - the installer build ran and seems to deliver an installable ppc64 tree 14:44:46 <kbsingh> however that is being done completely out of india / china / japan at the oment, but a group of folks who are not really co-ordinating too well amongst each other 14:45:05 <kbsingh> I've been trying to get heads together and set things up, but key requirement for that is to have a central build and host capacity 14:45:26 <MerlinTHP> Evolution and I are giving each other mutual (a)moral support on the armv7hl/aarch64 builds. 14:45:34 <kbsingh> been workin on that for a few months, we've had contract issues with redhat and the hosting company around space, but we should be close to closing on that 14:46:25 <kbsingh> given taht armv7 and powerpc are happening completely community side, it might be worth scoping up cbs involvement around it 14:46:34 <kbsingh> s/taht/that/ 14:46:53 <kbsingh> we dont need to do anything right now, but i wanted to bring it up, so its something that folks are thinking around / about :) 14:47:16 <MerlinTHP> I do have a bunch of questions around the arm stuff, like git, central build infrastructure, and whatnot. 14:47:48 <kbsingh> mikem: do you want to work through the 16th session, and see if that model works ? 14:47:53 <kbsingh> erm 14:47:55 <kbsingh> MerlinTHP: ^ 14:48:05 <MerlinTHP> Sure 14:48:09 <kbsingh> i *think* we can largely treat it as a SIG process 14:49:07 <MerlinTHP> As much fun as running a build infra out of my spare room is, I'd like us to get something a tad more formal if we're blessing it as centos ;) 14:49:14 <kbsingh> yeah 14:49:51 <kbsingh> i have a 4 way calxeda machine heading to the DC, ETA 10th Mar, that should give us 16 threads for armv7 ( these are 4 way cards each, i believe medway A15 ) 14:49:52 <bstinson> once we get the hardware, is it just process type things we need to iron out? 14:50:22 <MerlinTHP> kbsingh: is that the same Bostom Computing hardware that Fedora uses? 14:50:30 <kbsingh> bstinson: i think also aronud how we involve community, how we extend commnity ( hey, epel might want a armv7 build based off what we do etc ) - and how we message, deliver, set expectations etc 14:50:47 <MerlinTHP> Yeah, I was pondering what to do about epel 14:51:02 <MerlinTHP> If nothing else, _I'll_ want various epel packages for arm ;) 14:51:06 <kbsingh> MerlinTHP: this specific one isnt, but we have a boston box sitting in the UK waiting for the RedHat contract stuff to get sorted out, so we can get it into a DC here 14:51:13 <MerlinTHP> Ah, ok 14:51:31 <Evolution> for the epel stuff, the i386 agreement was to do it as a secondary arch 14:51:50 <Evolution> in theory we could do similar with armv7/aarch64 assuming they have the build boxen 14:52:29 <MerlinTHP> I was also wondering if maybe we should also look at armv6 14:52:40 <MerlinTHP> The redsleeve guys got that working 14:53:04 <MerlinTHP> I _think_ an armv7hl box should be able to build v6 in a mock chroot, similar to i686-on-x86_64. 14:53:08 <kbsingh> if there is room/scope for this - do we want to schdule a arm 32bit specific meeting ? 14:53:34 <MerlinTHP> v6'll be something to do once we have v7hl largely sorted 14:53:47 <MerlinTHP> kbsingh: do we want to roll aarch64 in too? 14:53:58 <MerlinTHP> Evolution's work is quite similar in a lot of regards. 14:54:33 <Arrfab> MerlinTHP: I have one armv6hl rebuilding (slowly) C7 SRPMS through mock right now 14:54:44 <MerlinTHP> Oh, awesome :) 14:54:59 <MerlinTHP> You also may want http://www.merlinthp.org/arm/notes.txt, then 14:55:09 <MerlinTHP> Hm 14:55:22 <MerlinTHP> In that case, we might want to look at sorting git stuff out sooner than later 14:55:25 <Evolution> reminds me. I need to write up an arm blog 14:55:26 <Arrfab> MerlinTHP: thanks, bookmarked, as well as the redsleeve public patches around amv6hl 14:55:45 <MerlinTHP> I've had to patch a handful of packages. 14:55:55 <MerlinTHP> I guess those changes really should go into git. 14:56:10 <Arrfab> MerlinTHP: I'm rebuilding against pidora19 atm 14:56:15 <MerlinTHP> Nice 14:56:42 <kbsingh> i need to rebase to another call, back in a bit. 14:57:12 <MerlinTHP> Arrfab: you shouldn't need to patch firefox and xulrunner btw, I don't think those NEON issues impact v6. 14:57:25 <bstinson> we're coming up on time for the CBS meeting, i'd like to close the minutes but i'll be around if we want to talk through some of this stuff 14:57:38 <Arrfab> bstinson: wfm 14:57:41 <MerlinTHP> Right, just one more quick thing, then 14:58:04 <MerlinTHP> Is there any chance we could configure a cgit instance on git.centos.org, as a faster read-only view of the git repos? 14:58:35 <MerlinTHP> For just doing read-only looking-at-git-repos, gitblit isn't great. 14:59:15 <MerlinTHP> tbh, I've mirrored the git repos locally and pointed a cgit install at it, but I suspect I might not be the only person who'd appreciate it on git.centos.org. 14:59:35 <kbsingh> cant ues cgit, it does not handle the underlyaing acl's 14:59:49 <kbsingh> underlaying even damn 14:59:50 <MerlinTHP> Is that actually an issue for read-only? 14:59:55 <kbsingh> yup 15:00:02 <kbsingh> there are private repos there as well 15:00:09 <MerlinTHP> OK, so filter those out. 15:00:25 <kbsingh> and some repos have private branches 15:00:30 <MerlinTHP> Or are we talking private... ah ok 15:00:31 <MerlinTHP> Hrm 15:00:34 <MerlinTHP> Feh. 15:00:41 * MerlinTHP sticks to a local mirror, then. 15:00:43 <MerlinTHP> OK, cheers. 15:00:53 <bstinson> :) 15:00:57 <kbsingh> but the newer gitblit is much faster, and has better url handling as well, so that will help 15:01:34 <MerlinTHP> And doesn't have the ///// bug, one hopes. 15:02:11 <kbsingh> yup 15:02:50 <bstinson> #endmeeting