fedora-meeting-1
LOGS
21:00:26 <pwhalen> #startmeeting Fedora ARM weekly status meeting
21:00:26 <zodbot> Meeting started Wed Feb  6 21:00:26 2013 UTC.  The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:26 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:00:26 <pwhalen> #chair pwhalen jonmasters bconoboy ctyler pbrobinson dgilmore
21:00:26 <zodbot> Current chairs: bconoboy ctyler dgilmore jonmasters pbrobinson pwhalen
21:00:32 <pwhalen> .fas pwhalen
21:00:33 <zodbot> pwhalen: pwhalen 'Paul Whalen' <pwhalen@redhat.com>
21:00:44 <masta> hello
21:00:44 <pwhalen> good afternoon all
21:00:46 <dmarlin> .fas dmarlin
21:00:47 <zodbot> dmarlin: dmarlin 'David A. Marlin' <dmarlin@redhat.com>
21:00:48 <masta> .fas parasense
21:00:50 <zodbot> masta: parasense 'Jon Disnard' <jdisnard@gmail.com>
21:00:51 <jcapik> .fas jcapik
21:00:53 <zodbot> jcapik: jcapik 'Jaromír Cápík' <jcapik@redhat.com>
21:01:15 * j_dulaney waves
21:01:38 <bconoboy> .fas blc@
21:01:39 <zodbot> bconoboy: blc '' <blc@redhat.com>
21:02:01 <ctyler> .fas chris@tylers.info
21:02:02 * ctyler tips his hat, sits down at the third seat on the left side of the conference table with a large Tim Horton's double-double.
21:02:04 <zodbot> ctyler: ctyler 'Chris Tyler' <chris@tylers.info>
21:02:11 <pwhalen> #topic 0) Status of ACTION items from our previous meeting
21:02:32 <bconoboy> what'd we agree to last week? :-)
21:02:35 <pwhalen> #link action items - http://meetbot.fedoraproject.org/fedora-meeting-1/2013-01-23/fedora-meeting-1.2013-01-23-21.01.html
21:03:07 <pwhalen> I've pinged everyone, looks like its pretty much done or on the agenda this week
21:03:23 <bconoboy> shall we record what was completed or not completed?
21:03:32 <bconoboy> I'm thinking anything still pending we want to log so it doesn't get lost
21:03:43 <bconoboy> (llvm jumps out at me)
21:04:20 * ctyler thinks llvm tends to do that
21:04:30 <pwhalen> agreed, pending would llvm, mongodb
21:04:57 <pwhalen> for myself, I need to organize a meeting to get the documentation from fudcon going
21:05:54 <pwhalen> masta, did you want to update us on ltrace?
21:06:37 <masta> hey there, unfortunatly I have not tested Petr'slatest hotness, but supposedly it should be working
21:06:59 <masta> we realyl want that tool to help troubleshoot issues in ARM, things liekthe recent polkit
21:07:03 <bconoboy> #action masta to test new ltrace package with potential arm fixes
21:07:09 <masta> I'll get a mock build done tonight and run some tests
21:07:25 <bconoboy> #action jonmasters still on the hook for mongodb and an llvm update
21:08:29 <bconoboy> anything else?
21:08:58 <pwhalen> I'm adding the 'secret decoder ring' document jsmith put together to the end of the meeting (time permitting)
21:09:21 <pwhalen> Peter and I spoke about the package list, there isn't one currently, but we'll revisit that
21:09:35 <pwhalen> nothing from me
21:09:42 <seano> a list of broken packages?
21:09:48 <bconoboy> #action pwhalen to work on 'secret decoder ring' document
21:10:03 <bconoboy> #action pwhalen & pbrobinson to continue work on broken package list (not yet available)
21:10:14 <pwhalen> I would like some input on that, as well as enroll some help with the pages we're to create, the list is long
21:10:47 <pwhalen> alright, next?
21:10:52 <seano> A lot of the stuff I saw, I had seen a lot of it previously on the website..
21:11:14 <seano> just not all in one spot..
21:11:37 <pwhalen> that encouraging, I do imagine it needs a refresh as well..
21:11:45 <pwhalen> #topic 1) Problem packages
21:12:12 <bconoboy> There appear to still be issues with eclipse and webkitgtk3
21:12:46 <bconoboy> There is also a bug in glibc that codonell is working on that results in some shared libraries not being found
21:13:14 <bconoboy> https://bugzilla.redhat.com/show_bug.cgi?id=905184
21:13:31 <bconoboy> #link Bug in glibc preventing ld.so.cache from working correctly https://bugzilla.redhat.com/show_bug.cgi?id=905184
21:13:48 <bconoboy> we should have a fix later today- we'll want to an updated f18 glibc for this
21:14:12 <bconoboy> anything that relies on ld.so.conf.d but hasn't been rebuilt since Halloween is affected
21:14:42 <pwhalen> afaik Peter needed to test the patch for webkitgtk3, he had some concerns the issue had not been fixed as the upstream bug was still open, and enabling could causing crashing rather then slowness
21:14:46 <pwhalen> but it was on his todo list
21:15:03 <bconoboy> for webkitgtk3 there's two possible solutions:
21:15:20 <bconoboy> 1. Turn JIT on.  It compiles with JIT on.  But we don't know how stable it is.  That would require testing
21:15:42 <bconoboy> 2. Take the patch proposed by the Fedora PPC team.  It will also allow the package to compile, and leaves the JIT off
21:16:04 <bconoboy> (We know it works without JIT)
21:16:22 <pwhalen> #2 sounds good for now, until we test #1
21:16:35 <seano> the problem with the patch is stability, or just not accepted upstream yet?
21:16:41 <ctyler> Do we have reason to suspect the JIT doesn't work? There's potentially a significant performance difference.
21:17:46 <pwhalen> historically it hasnt, caused crashes iiuc
21:18:18 <bconoboy> It crashed with gmail back in the day when JIT was turned on
21:18:30 <bconoboy> It's really time to test it again
21:18:35 <bconoboy> perhaps we could get a volunteer
21:18:51 <pwhalen> sure!
21:18:56 <ctyler> Is there a reduced test case or test plan available?
21:19:21 <seano> one that can add jazelle support to jit too? :)
21:19:42 <ctyler> Jazelle is dead...
21:19:49 <bconoboy> I'll provide a jit-enabled webkitgtk3 for armv7l to anybody who wants one.  You'll have to run a web browser that uses this library, then visit some javascript heavy websites and see what happens when you use them.  Gmail is a good test.
21:20:00 <seano> i thought it was included int he A15's even..
21:20:06 <pwhalen> bconoboy, sounds good, I'll take a look
21:20:20 <bconoboy> #action bconoboy to provide pwhalen with JIT-enabled webkitgtk3
21:20:36 <bconoboy> #action pwhalen to test JIT-enabled webkitgtk3 library on javascript-heavy web sites
21:21:32 <pwhalen> and eclipse?
21:21:45 <pwhalen> is that being looked at?
21:21:50 <bconoboy> just got an update on that
21:22:08 <bconoboy> #link https://bugzilla.redhat.com/show_bug.cgi?id=863801 should fix eclipse build issue
21:22:20 <bconoboy> so, that's it!
21:22:27 <bconoboy> (from me anyway)
21:22:29 <seano> ctyler: jazelle is in the a15 multicores..
21:22:38 <pwhalen> excellent, anything else?
21:22:56 <pwhalen> #topic 2) Wiki Gardening - removing outdated content
21:23:28 <bconoboy> Many wiki pages reference Fedora 17 being the latest release
21:23:34 <ctyler> saeno: the armv7-A profile doesn't require it afaik, the attention shifted to thumbEE
21:23:42 <bconoboy> This is no longer accurate
21:23:44 <pwhalen> in general it would be great if folks could go over our wiki, remove outdated content, point to f18 instead of previous releases
21:24:16 <ctyler> I noticed that some of the PA pages use a template for current release -- should we be doing the same?
21:24:22 <bconoboy> pwhalen: Perhaps a vfad day for this would be in order, make sure we set aside some time for it
21:24:45 <pwhalen> bconoboy, agreed, so we should also talk about the content additions suggested by jsmith
21:24:52 <pwhalen> as we discussed at fudcon
21:25:09 <seano> should probably be done at the same time..
21:25:16 <seano> or first..
21:25:29 <pwhalen> #link "secret decoder ring" doc - http://piratepad.net/L1jH2TLCBO
21:25:31 <seano> along with the secret decoder ring page..
21:25:53 <pwhalen> I'd love to set something up, divide and conquer the pages
21:26:07 <pwhalen> at the same time, plucking the outdated content
21:26:45 <bconoboy> #action s/17/18 on those fedora-arm wiki pages, everybody
21:27:06 <bconoboy> yes, much plucking - we want to move as much as we can to the main fedora wiki
21:27:10 <pwhalen> who's on board for some documentation vfad'in? and when would we like to do it ?
21:28:10 * ctyler is in if it's next week
21:28:19 * pwhalen hears crickets
21:29:04 <seano> Next week works better for me..
21:29:14 * jcapik says me
21:29:16 * bconoboy is in
21:29:20 <pwhalen> how does monday look then?
21:29:26 * dmarlin will help
21:29:30 <bconoboy> mondays are always awful, let's do it!
21:29:36 <pwhalen> :) done
21:29:37 <seano> is there a way to get a list of -all the pages?
21:29:53 <bconoboy> #action pwhalen to organize wiki vfad on monday
21:29:57 <pwhalen> click on the ARM category at the bottom, should list it all
21:30:23 <seano> should we just make a new page, and have all the pages listed and we can check off which ones were loooked it?
21:30:32 <pwhalen> *if* the category was added to the page when it was made
21:31:04 <bconoboy> seano: Let's just call out what page we're working on in channel on monday
21:31:16 <pwhalen> sounds like a reasonable idea, also look over the content to see if its needed, needs refresh etc
21:31:28 <seano> we are still going to need to keep track of it.
21:31:42 <seano> then write a bot script :)
21:32:27 <pwhalen> alright, next?
21:32:55 <pwhalen> #topic 3) Mass Rebuild - scheduling for ARM
21:33:31 <pwhalen> I think we may be missing the key folks for this one...
21:33:51 <ctyler> What's the latest word on (a) PHX2 and (b) PA mass rebuild?
21:34:05 <pwhalen> hub and db are ready to go
21:34:22 <bconoboy> PHX2: ARM hardware has arrived, but isn't racked yet
21:34:29 <pwhalen> the builders last I heard were going to be racked and a note dropped in channel as to when that was completed
21:34:29 <bconoboy> There is a ticket open for it to be racked.  No ETA yet.
21:35:43 <pwhalen> anything to be discussed?
21:36:17 <seano> if i lived close to pheonix i would go do it lol :)
21:36:23 <ctyler> >>crickets<<
21:36:32 <bconoboy> #info waiting on builders to be racked in PHX
21:36:34 * nirik continues to bug people to get it racked/networked. ;)
21:36:52 <bconoboy> nirik: is there any way to speed things up?
21:36:53 <pwhalen> thanks!
21:36:54 <masta> I thought hte mass rebuild was dellayed so those could be racked
21:36:59 <masta> *the
21:37:03 <bconoboy> is that why it was delayed?
21:37:20 <seano> that is kind of what i thought too.
21:37:20 * dmarlin thinks unlikely
21:37:27 <nirik> bconoboy: not sure. There's other work ongoing out there, so most of the onsite people are doing that... I'm trying to get as much info as I can.
21:37:28 <seano> although I didn't know it was delayed.
21:37:35 <nirik> one of the reasons, yeah
21:37:45 <ctyler> well, dgilmore is also the one who usually drives the MRs
21:37:54 <masta> exactly
21:38:12 <seano> i heard dgilmore had to go home on family biz.
21:38:24 <nirik> we were hoping it could be ready in time for primary so they could both be done at the same time. ;)
21:38:27 <seano> but that was 3rd party from an irc channel
21:38:29 <nirik> but we will see.
21:38:35 <masta> he's been silent lately
21:38:49 <ctyler> masta: yes, he's back in AUS
21:39:00 <bconoboy> nirik: who is going to drive the mass rebuild?
21:39:12 <nirik> dgilmore.
21:39:15 <bconoboy> okay
21:39:27 <seano> does he have a co-pilot?
21:39:37 <bconoboy> should we have contingency builders setup in HSV if the PHX stuff doesn't get racked in time?
21:39:42 * nirik will be happy to help too.
21:39:58 <pwhalen> that may be a good idea
21:40:10 <nirik> possibly. I still hold out hope we can get the new stuff in place, but time is getting low.
21:40:10 <bconoboy> We can throw a dozen or two hosts at it if needed
21:40:16 <seano> I think we need a copilot.. and a continegency plan..
21:40:39 <nirik> ok. I think the hub/db are all set, so we could use remote builders if need be
21:41:34 <ctyler> we'd need to cut over to the PHX2 hub/db, a point the HSV builders at it
21:41:42 <ctyler> a/ a / and /
21:42:20 <nirik> yeah.
21:42:50 <bconoboy> The actual transition from seneca to phx2 for hub/db is goign to be tricky- need dgilmore for that too, no?
21:43:55 <nirik> yeah, likely so.
21:44:05 <nirik> he's out today, but should be back tomorrow.
21:44:13 <nirik> (his tomorrow, our later tonight)
21:44:15 <ctyler> There are some pieces that are still tribal knowledge, like the updating of the user/package lists (which is pushed from somewhere in infra), so doing it without Dennis would likely be a slow go.
21:44:32 * ctyler would really like to see everything documented
21:44:36 <pwhalen> isnt that all in the db?
21:44:49 <nirik> that script lives in fedora infrastructure, I can change that. it just points to the dns for it tho I think.
21:45:07 <nirik> there's a script that syncs that info to all the koji hubs.
21:45:07 <ctyler> pwhalen: there's a push from FAS2 or somewhere on a regular basis to update that data
21:45:32 <seano> it needs to be documented
21:45:45 <nirik> it should just change anytime dns is changed.
21:46:14 <seano> you can hardset the dns to see if it works.
21:46:40 <nirik> I wouldn't want to change it until we change the hub
21:47:12 <seano> i meant to test just change it in /etc/hosts.
21:47:18 <nirik> it just uses 'arm.koji.fedoraproject.org', so when dns changes it should follow fine.
21:47:25 <bconoboy> I think we need to ask dgilmore to send an update to the list when he's free
21:47:31 <nirik> sure.
21:48:02 <bconoboy> #info Red Hat can provide auxiliary builders if needed to handle mass rbuild
21:48:08 <nirik> the big things on moving hub would be: stop all builds/shutdown current hub, db dump/import? then change dns.
21:48:34 <pwhalen> final rsync on mnt/koji
21:48:45 <bconoboy> #info dgilmore to manage koji transition (please send email with plan to list)
21:49:07 * nirik nods, yeah.
21:49:27 <pwhalen> alright, shall we move on?
21:49:47 <pwhalen> #topic 4) Your topic here!
21:50:07 <pwhalen> thanks for the update nirik
21:51:39 <bconoboy> I have one
21:51:48 <bconoboy> Fedora 18 is out- You are all awesome for chipping in.  Thank you!
21:52:05 <pwhalen> our new 3.7 kernels lack the audit config changes, when your f18 system is updated, you'll lose the fix. I let Peter know, new build is pending
21:52:24 <bconoboy> #info Current 3.7 kernel shows pkexec regression- new build is pending
21:52:44 <masta> it's ok
21:52:50 <bconoboy> #info F18 ARM is out- #fedora-arm victory lap!
21:52:57 <pwhalen> \o/
21:53:05 * masta high fives everyone in the team
21:53:49 <pwhalen> anything of note on the release? anyone have issues?
21:54:15 <seano> im having network issues
21:54:17 <seano> :)
21:54:21 <pwhalen> we discussed changing the fstab on the rootfs tarball, not sure if everyone saw
21:54:29 <pwhalen> seano, which image? what happens?
21:54:31 <bconoboy> did we get resolution?
21:54:38 <pwhalen> we didnt
21:54:47 <seano> armv5, and i also had some weird reversion after reboot
21:56:26 <seano> i either have media problems, or there is something funky with I would guess shutdown and journaling..
21:56:27 <masta> the rootfs tarball should have LABEL=rootfs, as a generic
21:56:39 <bconoboy> +1
21:56:56 <bconoboy> anything for /boot or /boot/uboot should probably be removed too.  They're not necessarily needed
21:57:21 <pwhalen_> sorry, I dropped and missed the last comment (if any)
21:58:02 <masta> pwhalen_: removing all fstab entries except rootfs
21:58:14 <pwhalen_> so change to use a label?
21:58:33 <dmarlin> so what happens when the internal drive/ssd is using  LABEL=rootfs ?
21:58:45 <dmarlin> conflict
21:58:47 <seano> dmarlin you get a mix up. :)
21:58:53 <dmarlin> *nod*
21:59:02 <dmarlin> bad choice, IMO
21:59:04 <bconoboy> how about LABEL=remixrootfs
21:59:13 <masta> pwhalen_: I'd say so, it's the most generic thing to have... of course the consumers of this tarball will have to make their own changes....
21:59:31 <dmarlin> bconoboy: why not just document what people need to do to use these, including adding an fstab?
21:59:33 <masta> bconoboy: +1
21:59:36 <pwhalen_> I think, we should leave it as is and document it
21:59:42 <seano> that should just be a documented thing..
21:59:45 <bconoboy> or we could make it a broken fstab file, don't say LABEL=anything, instead say "ROOTDEVGOESHERE"
21:59:45 <masta> we are just sending a msg to the remix creator
21:59:54 <seano> renaming rootfs might cause other issues.
22:00:05 <pwhalen_> reason being is, it will need to be changed, and wont require additional scripting on the tool
22:00:19 <masta> sure
22:00:33 <bconoboy> dmarlin: that'd work too.  It looks like pty and stuff doesn't there any longer
22:00:49 <dmarlin> bconoboy: right
22:01:00 <ctyler> Do we even need a root entry?
22:01:13 * dmarlin does not think so... not to boot
22:01:17 <masta> ok folks, got to run off now... will read the logs for the ending
22:01:21 <seano> i think to shutdown you do
22:01:32 <ctyler> seano: to shutdown??
22:01:36 <dmarlin> ?
22:01:44 <seano> umounting filesystems..
22:02:08 <ctyler> well, if you can shutdown with a bad fstab, then it wouldn't seem to matter
22:02:33 <seano> i can't remember if it will shutdown cleanly.
22:03:48 <seano> and it seems like the kernel reads it when it remounts rw.. because you can add mounting options there
22:05:37 <pwhalen_> anything else folks?
22:06:49 <bconoboy> not I
22:08:17 <seano> not i..
22:08:19 <pwhalen_> #endmeeting
22:08:27 <bconoboy> #endmeeting