20:01:27 <dgilmore> #startmeeting
20:01:27 <zodbot> Meeting started Fri Dec 10 20:01:27 2010 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:01:27 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:01:36 <dgilmore> #name fedora releng
20:02:06 <dgilmore> #meetingname fedora releng
20:02:06 <zodbot> The meeting name has been set to 'fedora_releng'
20:02:15 <dgilmore> #topic roll call
20:02:20 <dgilmore> ping all
20:02:29 * nirik is lurking around.
20:02:57 * notting is here
20:04:09 <dgilmore> Oxf13: spot: skvidal:
20:04:16 <dgilmore> abadger1999:
20:04:23 <skvidal> oh
20:04:25 * skvidal is around
20:04:38 * abadger1999 somewhat around.
20:04:40 <Oxf13> I won't be here, gotta run some errands
20:04:44 <dgilmore> anyone ok
20:04:54 <dgilmore> well the agrenda i forgot to send out is light
20:05:16 <dgilmore> #info present skvidal, notting, abadger1999, nirik
20:05:27 <dgilmore> #info also present dgilmore
20:05:33 * jsmith is lurking
20:05:45 <dgilmore> #topic cloud sig follow up
20:06:02 <dgilmore> so i dropped in on the cloud sig meeting yesterday
20:06:59 <dgilmore> the result of that is that they will provide a kickstart and get it into spins-kickstart git repo for image composition
20:07:28 <nirik> cool
20:07:40 <dgilmore> we will make amis
20:07:43 <jsmith> Should be a fairly simple kickstart, too :-)
20:07:57 <dgilmore> and use a modified verison of rhels upload scripts to upload the results
20:07:58 * jds2001 is lurking too
20:08:10 <nirik> is someone from that group going to maintain it? I can add them to the group to commit there...
20:08:24 <dgilmore> jgreguske will be helping with the upload bit
20:08:35 <nirik> ok.
20:08:36 <dgilmore> nirik: likely jforbes will
20:09:35 <dgilmore> as part of building livecds in koji we will build the appliances also
20:10:56 <dgilmore> anyone else have any infput?
20:10:59 <dgilmore> input
20:11:14 <nirik> sounds reasonable to me. Is this just one image? or ?
20:11:27 <dgilmore> just one image
20:11:33 * jds2001 is unsure what this image is or is for
20:11:39 <jds2001> is it the EC2 image?
20:11:42 <dgilmore> its a pretty minimal one from what ive seen
20:11:45 <dgilmore> jds2001: yep
20:11:52 <dgilmore> the EC2 image
20:12:16 <dgilmore> ideally we could upload the one image to amazon, rackspace, other cloud providers
20:12:33 <jds2001> would be nice :)
20:12:34 <dgilmore> or we just publish it and they can pull it down and install it
20:12:58 <dgilmore> i guess we will need to communicate with the cloud providers
20:13:01 <notting> well, different images for different providers, alas
20:13:03 <dgilmore> and work out how to do it
20:13:09 <jds2001> so how does this differ from the AOS image?
20:13:15 <jds2001> if it's intended to be generic?
20:13:28 <notting> is the aos spin still maintained?
20:13:33 <jds2001> no clue
20:13:46 <dgilmore> AFAIK the aos spin is no longer maintained
20:13:56 <nirik> it still exists, but I don't think we have produced it in several cycles.
20:14:04 <jsmith> dgilmore: Actually, there may two images (i386 and x86_64)
20:14:16 <dgilmore> jsmith: right 2 arches
20:14:21 <dgilmore> but the same content
20:14:26 <jsmith> dgilmore: Just wanted to be clear :-)
20:14:49 <jsmith> dgilmore: There may also be a need for an S3 backed image and an EBS backed image -- more discussion to happen on the cloud list
20:14:52 <jds2001> not many commits lately in aos :(
20:15:10 <jds2001> er, you mean local instance and EBS?
20:15:15 <dgilmore> jsmith: ok
20:15:32 <jds2001> but i think the image is the same, it's just the AMI definition that differs.
20:15:43 <jds2001> anyhow, that's for the cloud SIG meeting :)
20:16:21 <dgilmore> ok anything more?
20:16:44 <dgilmore> #info cloud sig to provide kickstart for EC2 image composition
20:17:25 <dgilmore> #info dgilmore and jgreguske to work on scripts for uploading the resulting images
20:17:53 <dgilmore> #topic open floor
20:18:06 <dgilmore> any one have anything that they want to discuss?
20:18:26 <skvidal> dgilmore: oh the tcket is here: https://issues.jboss.org/browse/BGBUILD-73
20:19:00 <abadger1999> I added the intro job information and little blurb on the join page we talked about last time
20:19:03 <nirik> I have one controversal item.
20:19:07 <dgilmore> skvidal: that ticket doesnt matter
20:19:13 <dgilmore> we are not using boxgrinder
20:19:22 <dgilmore> nirik: ok
20:19:25 <skvidal> ok
20:19:35 <abadger1999> Would also be good to get a rel-eng orientation page similar to infra's.
20:19:40 <notting> dgilmore: we are using ... koji?
20:19:49 <dgilmore> notting: yep
20:20:03 <dgilmore> kojis appliance building functionality
20:20:15 <nirik> Not sure if this is a rel-eng/fesco/whatever question, but how much yelling and invective would I get for asking that lxde/xfce groups be added to the dvd? ;) They are pretty small...
20:20:26 <dgilmore> one day that may be done by koji/boxgrinder integration
20:20:43 <dgilmore> nirik: id be ok with it
20:20:55 <jds2001> does it make it go over size?
20:21:05 <dgilmore> nirik: we probably would want to do a compose and see how big it is
20:21:31 <nirik> jds2001: there's lots of space last I looked... but I can investigate more if people don't just hate the idea off hand. ;)
20:21:59 <Southern_Gentlem> ? and add fpaste to the livecd please
20:22:05 * jds2001 is an equal opportunity desktop user :)
20:22:31 <jds2001> Southern_Gentlem: what does that pull in? anything?
20:22:42 <notting> nirik: well, i dislike the DVD in general, but don't necessarily have a problem with that
20:22:42 <Southern_Gentlem> nope
20:22:46 <nirik> it wouldn't show up or anything unless someone did a 'customize now' and choose it, etc.
20:23:01 <nirik> well, fpaste pulls xsel, but it's also small.
20:23:21 <jds2001> yeah, i just looked and found that, both are tiny
20:23:32 <Southern_Gentlem> 100k
20:24:05 * jds2001 has no objections to fpaste
20:24:29 <jds2001> though the packaging is bad, since it doesnt require python, but that's a nit :)
20:24:32 * dgilmore wonders why
20:24:47 <jds2001> dgilmore: it helps for irc support
20:24:48 <dgilmore> but has no objection to it
20:24:54 <jds2001> cat file | fpaste is easier
20:25:09 <Southern_Gentlem> fpaste --sysinfo is easier
20:25:11 <nirik> but of course it won't be installed by default unless it's in a group that gets installed by default.
20:25:28 * jds2001 has it on my system and cant recall installing it
20:25:36 <jds2001> though maybe i did and didnt realize it :)
20:26:00 <dgilmore> jds2001: same
20:26:55 <jds2001> so now on to a controversial topic that i had.
20:27:01 <dgilmore> jds2001:  ok
20:27:07 <jds2001> and i can already here the screaming :)
20:27:14 <jds2001> s/here/hear
20:27:34 <jds2001> anyhow, i think the spins wrangling function belongs in releng
20:27:49 <jds2001> in order to facilitate, im willing to take about half the load :)
20:27:54 <dgilmore> spinds wrangling in what sense?
20:28:18 <jds2001> making sure spins are within size, that they compose successfully every night, etc.
20:28:39 <jds2001> general stuff. Spins are responsible for fixing issues when they arise
20:28:42 <dgilmore> jds2001: i disagree there
20:28:53 <dgilmore> at least as to the size bit
20:28:55 <jds2001> unless me or someone is feeling particularly charitable that day :)
20:29:18 <jds2001> why is size not a concern?
20:29:24 <dgilmore> i can see releng doing the composing as part of the rawhide compose
20:29:41 <nirik> we typically have not worried too much about size until betaish.
20:29:57 <dgilmore> but the people that care about a spin and are intimate with the software should be the ones making the decision to drop packages
20:30:45 <jds2001> note that i said the people that care need to fix it :)
20:31:11 * nirik is happy to keep doing nightly composes and such. If someone is willing to bug maintainers and get them to do things, that would be fine with me.
20:31:14 <jds2001> but my guess is folks dont pay attention.
20:31:37 <nirik> right.
20:31:49 * abadger1999 thinks if we're pushing spins uner the rel-eng umbrella, it would be best to get spin people into rel-eng to do the building.
20:32:00 <jds2001> abadger1999: exactly
20:32:35 <nirik> so each spin maintainer would need to join to build their stuff?
20:32:46 <dgilmore> abadger1999: well with livecd etc composition automated for releases we could do the same for nightly composes and integrate it into the rawhide process
20:32:48 <notting> jds2001: no, releng should not be responsible for making sure spins are within size. only for kicking SIGs when they aren't
20:33:12 <jds2001> notting: i think thats what i said in another way
20:33:17 <jds2001> releng pays attention and kicks :)
20:33:32 <abadger1999> dgilmore: Does spin composing ever break?
20:33:37 <jds2001> kicked party fixes :)
20:33:45 <dgilmore> abadger1999: in theory
20:33:45 <nirik> the problem with spins is not the composing or checking size or the like. It's testing, being on time for things, meeting QA, following deadlines, helping review new spins.
20:34:14 <nirik> abadger1999: spins often fail to compose for various amounts of time. Broken deps, package renames, etc.
20:35:42 <abadger1999> My impression is that there's two parties involved and things get dropped when it moves between them -- Spins create the kickstart.  Then rel-eng creates them if possible.  Then spins are supposed to test them.  rinse and repeat if necessary.  We get failures when the process transitions from one group to the other and the new group doesn't start on the next steps.
20:35:48 * notting queues an open floor question
20:36:50 <dgilmore> abadger1999: i wonder if we should only compose nightly livecds when we compose install trees also
20:36:55 <dgilmore> i.e not rawhide
20:36:57 <abadger1999> So if there's not the transition -- ie: the spin owner creates the spin and then tests it... it seems less prone to last minute, oh I didn't know the new images were available complaints.
20:37:01 <dgilmore> but branched
20:37:56 <dgilmore> i wonder if mirrors would bitch at new livecds getting uploaded nightly
20:37:56 <jds2001> abadger1999: im not sure how the livecd-in-koji thing works, but does that help that problem at all?
20:38:13 <jds2001> dgilmore: they're on alt today
20:38:38 <dgilmore> jds2001: yeah im thinking not on alt  but in the branched tree
20:39:11 <dgilmore> anyways we have some things to look at here
20:39:29 <dgilmore> #info look at integrating nightly livecds with branched composing
20:39:54 <dgilmore> #info look into ways to better test livecds
20:40:01 <dgilmore> lets move on for now
20:40:08 <notting> one item
20:40:21 <notting> apparently the dep report for EPEL 6 came out with  'rawhide' in it?
20:40:28 <dgilmore> #task jds2001 to work on integrating livecds
20:40:43 <dgilmore> notting: thats probably an oversight on my part
20:41:24 <dgilmore> notting: it was
20:42:52 <dgilmore> notting: i forgot to pass in a treename
20:42:59 <notting> just wanted to raise it so it can get fixed
20:43:08 <notting> seems to have confused a couple of people on-list
20:43:11 <dgilmore> notting: thanks fixed already
20:44:51 <dgilmore> anyone have anything else?
20:44:56 <dgilmore> if not ill close in 30
20:45:44 * jsmith has nothing else to add
20:49:38 <dgilmore> #endmeeting