20:00:57 #startmeeting 20:00:58 Meeting started Wed Sep 5 20:00:57 2012 UTC. The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:58 Useful Commands: #action #agreed #halp #info #idea #link #topic. 20:01:10 #chair pwhalen jonmasters bconoboy ctyler pbrobinson dgilmore dmarlin 20:01:10 Current chairs: bconoboy ctyler dgilmore dmarlin jonmasters pbrobinson pwhalen 20:01:24 * pbrobinson waves 20:01:34 this meetbot was made by debian? 20:01:39 .fas pwhalen 20:01:41 pwhalen: pwhalen 'Paul Whalen' 20:01:50 -> http://wiki.debian.org/MeetBot. ? 20:01:55 * masta is here 20:01:57 .fas ctyler@ 20:01:58 ctyler: 'ctyler@' Not Found! 20:02:02 fossjon: maybe, why does it matter? 20:02:09 it should be fedora only! 20:02:10 .fas ctyler@fedoraproject.org 20:02:13 ctyler: 'ctyler@fedoraproject.org' Not Found! 20:02:17 .fas ctyler 20:02:18 ctyler: ctyler 'Chris Tyler' - ctyler2621 'Christopher Tyler' 20:02:19 .fas Frojoe 20:02:21 Frojoe: burningfool 'Jordan Cwang' - jacwang 'Jordan Cwang' 20:02:29 .fas chris@tylers.info 20:02:30 ctyler: ctyler 'Chris Tyler' 20:03:05 * ctyler finally figures out .fas syntax to get just his entry 20:03:12 :) 20:03:21 anyone else expected to join? 20:03:23 why do we even need .fas entries anyway? 20:03:28 .fas Chr.*Tyl 20:03:29 fossjon: 'Chr.*Tyl' Not Found! 20:03:36 regex doesnt seem to be supported 20:03:38 hmm 20:03:45 we need a new and improved fedora bot 20:03:46 pbrobinson: Just a quick way to give more info about the attendees. 20:04:06 okay... 20:04:13 #topic 1) F18 Build status - problem packages? 20:04:13 pbrobinson: I did that in some other fedora meeting, was the only one.... akward 20:04:42 masta: haha :) 20:05:19 pbrobinson, any big blockers you want to call out? 20:07:04 glibc in f19 is our big blocker at the moment 20:07:22 none really of importance for f18 that are major blockers at the moment 20:07:43 glibc in f19 is the reason why out f19 numbers are up in the daily report 20:08:20 EOF 20:08:45 #link http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=1114826 20:09:13 #info glibc is currently the only major blocker in f19 20:09:32 ah, ghc is a blocker on f18 but it only affects haskell and one of the maintainers is aware of the issue 20:09:46 he's on hols and it's on his list for his return 20:10:10 we're not using haskell to actually build anything? 20:10:41 nothing of major importance 20:11:11 there's a lot of ghc* packages and it's blocking some of those but it will unblock pretty quickly when person is back 20:11:30 anything else on build status or blockers, anyone? 20:12:02 #topic 2) ARM Koji Storage - more storage needed 20:12:50 Ok, we added about 600G of SSD to /mnt/koji a short while back, and we've burned through it already 20:13:10 Wondering if someone can help me project needs over the next while 20:13:36 Which might be easier if we had a PHX target date in mind 20:13:43 does it still contain < F17 packages? 20:13:47 Have we been building that much that we ate through 600GB already? 20:13:55 and can we archive those somewhere / how 20:14:34 everything is inherited since f15 though right? 20:14:42 I suspect it does, but not sure that pre-F15 buys us much space 20:14:54 so we could drop anything older than that? 20:15:44 But even with dropping old stuff, it would be good to try and estimate needs until PHX. 20:15:51 I think we have f12/13/14 there and that might buy us enough space until we move to primary arch 20:16:17 Also, I'm hesitant to just move/nuke stuff without letting the Koji db know. 20:16:42 dgilmore would likely be the best to estimate and inform us of what or if we can archive. Not sure he's here today though 20:16:43 f12 was not built on this Koji instance, so it's just f13/14 20:17:14 Ok, I'll start a thread on the list and we can back-of-envelope there. 20:18:00 great 20:18:43 #info ctyler to send out item to the list to discuss estimated Arm Koji disk usage 20:19:05 #topic 3) Kernel testing - looking for volunteers for each platform to do regular testing of the upstream kernel 20:19:25 define "upstream kernel" 20:19:36 kernel.org 20:19:41 What kind of testing? 20:19:54 not sure what testing the upstream kernel gets us 20:20:08 pwhalen: Can you detail the full picture of what you want? 20:20:09 we barely have the resources to deal with the Fedora kernel 20:20:54 looking for someone for each platform to attempt to build the upstream kernel using our config, document the problems they run into 20:21:20 I think the theory is that if we keep upstream building/booting on each platform, then fedora will require less effort 20:21:31 so it's not really kernel.org kernel at all once it's had our config applied. Why not just a fedora rawhide kernel 20:22:05 because rawhide follows mainline _VERY_ closely anyway 20:22:06 pbrobinson: how close does that run to kernel.org (where the patches need to go) 20:22:24 pbrobinson, sure, would be a great starting point. As we know the kernel is always an issue for us, just identifying issues often and early is the hope 20:22:41 pbrobinson: if its close enough to submit patches against upstream, it sound fine to me 20:22:43 very, they pull every day or two and add a few other bits on occasion like secure boot or wireless updates 20:23:09 so we should be testing rawhide kernels regularly not mainline 20:23:18 pbrobinson: ok, so s/kernel.org/fedora rawhide/ 20:23:20 dmarlin: yes 20:23:45 same theory... just different source 20:24:09 We'd be glad to test across a group of boards here. 20:24:10 It's something that I've been trying to get people to do for quite some time. It's great that others are starting to agree 20:24:20 How "regularly" do you want to do this? 20:24:26 the job is rather large, the idea is to split it up in our group, have someone keeping as close an eye on one platform 20:24:28 * masta is willing for panda and tslice 20:24:34 every rcX would be useful IMO 20:24:48 if someone (or some group) with a vested interest in each platform would volunteer to 'own' that kernel and make sure it builds/boots, it would be very helpful... 20:24:58 and submit fixes upstream 20:25:00 so just build/install a rawhide kernel for each board? right? do we have a wiki spot to make reports, or what? 20:25:12 rc4 is steady building and includes a few arm fixes 20:25:30 I'm quite happy to push fixes into the fedora kernel. I would love some assistance in pushing to mainline kernel.org fixes. It's something I need to work out how best to do it 20:25:32 what do you want tested beyond "it boots!"? 20:25:38 masta, new idea, but wiki page to come 20:25:43 masta: I'd suggest just one kernel, as the job can still be quite large 20:26:11 hola 20:26:33 pbrobinson: good point... how best to get changes upstream to mainline 20:26:39 ok so let pretent masta uses rawhide kernel once a week, find an issue... then what? 20:26:58 ctyler: ive been told for primary we can not remove anything from koji thats in a release, either ga or update 20:27:00 I do have this kernel testing results page up - http://fedoraproject.org/wiki/Architectures/ARM/Quality_Assurance/Kernel_Testing 20:27:07 masta: or more often... every new build. 20:27:28 ctyler: if it boots it would be useful for eventually testing wider hardware features and see what works. Things like wireless, display, sound etc. possibly gpio. It's something we've not really done wide testing on 20:28:05 pbrobinson: agreed, but right now simply "not broken" would be an improvement 20:28:13 yes 20:28:23 hence I said "eventually" 20:28:50 agreed 20:28:53 Ok, so we'll do this whenever a kernel appears in ARM rawhide? 20:29:19 ctyler: I would think so. 20:29:43 we can start with that and adjust as necessary 20:29:46 ctyler, that sounds like a reasonable starting point, keep adding to the results 20:29:48 * ctyler pokes fossjon about doing notifications for this 20:30:02 can we do some form of summary to the mailing list 20:30:18 would also be useful to some how report regressions 20:30:34 i see 20:31:09 * masta would love to try new kernel hotness once or twice a week, would be great to have a QA check list of things to test 20:31:16 Best bet would be to run this kernel with $lastreleased userspace? Running with rawhide userspace would keep us guessing on whether it was kernel or userspace that broke stuff. 20:31:42 I'm quite happy for it to be run on an older known good userland 20:31:46 pbrobinson, thats a good idea, I'll look at sending something out on a regular basis. We have a new kernel nearing finish as well. rc4 has a few arm fixes 20:31:48 ctyler: that sounds like a good idea 20:32:10 pwhalen: I know it has :-P 20:32:51 okay, sounds like a good plan. for sfp any volunteers? 20:33:26 pwhalen: just to be clear, maybe recap the 'job description' ? 20:33:55 - build latest fedora rawhide kernel 20:34:04 - if fails, fix and submit patch 20:34:10 dmarlin: not build, that will come from koji 20:34:14 - if builds, test boot. 20:34:35 pbrobinson: what if it fails to build? 20:34:38 dmarlin: you'll get more takers for "test" than "fix and submit patches" 20:34:54 ctyler: yes, but "fix and submit patches" is what we need 20:35:01 but I think we're looking for both 20:35:20 agreed... but I was thinking someone who could do both 20:35:24 test, look for patches or tell everyone about it 20:35:32 dmarlin: to date only dgilmore and I with some help from pwhalen have ever really done upstream mainline fedora build fixes 20:35:38 and/or "scream about regression" 20:35:59 pbrobinson: right, and I was hoping some others here could help... spread the load 20:36:39 for a "fails to build" they could do diffs of the configs in /boot and some basic regression googling but I wouldn't in the short term expecting people to build new kernels etc 20:36:43 often required patches are but a google search away, and when their not, telling the list may help solve the issue 20:37:39 yeah, but a googled patch != acceptable upstream fix 20:38:21 ctyler: in a lot of cases the fix is in an ARM branch such as arm-soc ready for the next dev cycle, or the next rc pull so in a lot of cases they are 20:38:25 so, ideally we're looking for testers of successful builds, but also someone willing to take ownership of a specific platform when they also fail. to identify the issue 20:38:55 pwhalen: +1 20:39:27 in all cases in fact the fixes I'm put in have been 1) upstream somewhere, branch etc 2) a basic 1 line include fix 3) some Kconfig dep missing 20:40:37 pbrobinson: adding some lessons-learned on the wiki could help others new to the process 20:41:16 alright, perhaps we can either take this to the list, or those interested could contact myself. Not sure I feel many bites 20:41:16 yep, I think we need to setup a wiki page and I think the whole process will involve over time 20:41:21 pbrobinson: common branches to check, types of config issues 20:41:36 *nod* 20:41:40 We're glad to take on any platforms we have here (guru/sheeva, panda, beagle, trimslice, smarttop, oscar/felix) 20:42:03 ctyler, sounds like you got it covered! 20:42:12 I think a single wiki page covering people assigned to particular kernels and all so tips and process etc 20:42:31 pbrobinson: +1 20:42:38 +1 20:42:59 +1 20:43:07 pwhalen: You'll do wiki page + mail to list? 20:43:20 so who's going to kick off and initial wiki page and drop a mail to the list soliciting suckers ^w testers once they have it in place 20:43:52 ctyler, sure 20:43:55 ctyler: what is oscar/felix? 20:43:57 * ctyler pwhalen to create kernel-testing wiki page including test procedure, mail to the list 20:44:02 * masta will be one of the suckers 20:44:07 pbrobinson: oscar/felix are $NDA boxes 20:44:20 One soft, one hard. 20:44:22 #action pwhalen to create kernel-testing wiki page including test procedure, mail to the list 20:45:04 who is meeting chair? Is that you pwhalen? 20:45:23 #topic 4) Fedora ARM F18 Alpha release VFAD - was cancelled this week, when should(can) we reschedule for? 20:45:29 pbrobinson, I am 20:45:47 you need to do actions etc so we can get decent meeting notes! 20:45:49 #action pwhalen to create kernel-testing wiki page including test procedure, mail to the list 20:46:00 so, we cancelled the vfad. No booting kernel, no images. The pa release was also delayed 20:46:20 so we kicked the can down the road 20:46:22 pbrobinson is also a chair 20:46:34 How's PA doing? I saw some good reports on rc5 20:46:42 Haven't tried myself though 20:46:43 pwhalen: there are booting kernels. I believe vexpress and highbank in the very least 20:46:55 ctyler: you mean tc5? 20:46:56 vexpress does not find the rootfs 20:46:58 well, not *enough* booting kernels 20:47:10 yeah 20:47:18 highbank does boot 20:47:51 panda? 20:47:57 nope 20:47:59 http://fedoraproject.org/wiki/Architectures/ARM/Quality_Assurance/Kernel_Testing#kernel-3.6.0-0.rc3.git4.1.fc18 20:48:25 so chatting with dgilmore today we had a booting omap kernel with 3.5 but no usb with the current f18 config. It seems 3.6 has regressed there some how. It looks like it might be something to do with the clk framework stuff. 20:48:48 the usb issue will be resolved with the current building rc4 kernel 20:50:15 s/will/should/ :-) 20:50:35 so, we need to have booting kernels and the ability to produce the images prior to a vfad 20:50:48 ctyler: No. the usb will be solved. The boot issue I'm not so sure about 20:50:52 dmarlin, is going to look at creating the images 20:51:21 f18 images 20:51:24 so did we move away from using labels and go with uuid? 20:51:27 f17 already working 20:51:44 masta: the kickstarts I use do 20:51:52 our vfad plans are up in the air until those issues are resolved 20:52:00 anaconda will use UUID's 20:52:28 Is it reasonable to think we can be in a position for a VFAD in a week? 20:53:58 ctyler: what we really need is some scripts to generate our media. 20:54:08 that could be worked on without things working 20:54:11 I would hope so but I'm on holidays from next week and I have family visiting so I'll be about but won't have a lot of time to dedicate to kernel things 20:54:53 I think bconoboy mentioned reviving his scripts? 20:55:16 I can keep tabs on both and send something out to the list if we've been successful 20:55:17 masta: those are not appropriate 20:55:39 the scripts should call livemedia-creator to make all the images 20:55:54 what was used in the past was nothing at all like what we will use today 20:55:56 we really need to be moving forward to kickstarts 20:56:30 pbrobinson: indeed 20:56:31 we need to be looking forward to what's needed for PA promotion which is using mainline tools 20:56:44 the issue, iiuc is the creation of the images using f18 tools, dmarlin? 20:56:47 there is the arm branch of spins-kickstarts 20:56:51 that should be used 20:57:32 pwhalen: I'm just starting to try f18 tools. the patches are upstream, but I need to set up an f18 chroot to test in 20:58:55 okay, so I'll keep in the loop with dmarlin, watch for kernels that boot, and if we have these items by next week we have a alpha vfad? 20:58:56 dmarlin: what's the timeframe for that testing to happen? 20:59:39 pbrobinson: I'm working on it (between other things)... 20:59:40 Hour marker is approaching 20:59:48 pbrobinson: I hope to have something working this week 21:00:35 great 21:00:43 what else do we need to cover. I need dinner! 21:01:02 #info pwhalen to notify the list if image creation is success using F18 tools, F18 kernel is built and we are ready for a VFAD next week 21:01:12 #topic 5) Your topic here 21:01:17 Two things: 21:01:19 (1) I'd like to introduce Dmitry Kozunov aka DarthJava. 21:01:20 Dmitry will be working full-time with the Fedora ARM team here through the end of the year and will be a good contact for issues with the buildsystem. 21:01:21 (2) I have a class of students learning about packaging and build & release and need some projects for them. Ideally, projects that we're 21:01:22 interested in having done, have a reasonable scope (3 months of part-time hacking), and are not critical-path/blockers -- in other words, those 21:01:24 thats all I had 21:01:24 things we have on our "if only I had some spare time" lists. Best if they're single-person scope and/or can be split into single-person 21:01:26 pieces, and if you're willing to provide the ocassional pointer or piece of advice. Please mail/IRC me if you have some ideas. 21:01:28 ouch 21:01:52 EOF 21:01:54 welcome DarthJava 21:02:06 thank you 21:02:36 greets DarthJava 21:02:44 :) 21:03:09 ctyler: so does DarthJava replace maxam? 21:03:20 pbrobinson: Correct. 21:03:32 Also, fossjon is full-time over the next year :-) 21:03:41 hooray! 21:03:45 the fedora arm team : DarthJava agreene Frojoe fossjon ctyler 21:03:46 agreene and Frojoe are part time. 21:03:58 ctyler: i have some ideas on projects, do you have a time frame you need details by? 21:04:18 pbrobinson: ideas over the next couple weeks would be awesome 21:04:27 they'll start in about a month 21:04:56 anything else, anyone? 21:05:07 ctyler: I'll try and throw some stuff into a mail to give a start and we can go back and forward as necessary. Quite happy to help where I can 21:05:19 pbrobinson: thanks :-) 21:05:29 * masta is helping the ambassadors setup and promote fedora-arm at events 21:05:36 #endmeeting