16:29:17 <dgilmore> #startmeeting
16:29:17 <zodbot> Meeting started Mon Jan  6 16:29:17 2014 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:29:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:29:23 <dgilmore> #meetingname releng
16:29:23 <zodbot> The meeting name has been set to 'releng'
16:29:34 <dgilmore> #topic whos is here
16:30:01 <dgilmore> ping nirik masta dwa tyll_ rdieter
16:30:09 <nirik> hey.
16:30:32 * dgilmore failed to send out an agenda
16:30:44 <dwa> morning
16:30:47 <tilldroid> I am on  my way
16:30:53 <danofsatx> I'm here, but not "officially" releng (yet)
16:31:43 <tilldroid> I am tyll
16:31:46 <dgilmore> danofsatx: welcome :)
16:31:54 <dgilmore> tilldroid: :)
16:32:11 * rdieter waves, hola
16:32:21 <dgilmore> will wait a couple of minutes for dwa and masta
16:32:21 * sharkcz is here
16:32:25 <dgilmore> hola sharkcz
16:32:30 <sharkcz> hi Dennis
16:32:43 <dwa> dgilmore: I'm here
16:32:57 <dgilmore> sorry i forgot to ping you sharkcz
16:33:37 <sharkcz> np :-)
16:36:28 <dgilmore> okay lets get started
16:36:55 <dgilmore> #topic tickets
16:37:07 <dgilmore> right now we have no tickets in the meeting queue
16:37:41 <dgilmore> #info
16:37:51 <tilldroid> There are some old Tickets wäre it is unclear what the next action is
16:37:53 <dgilmore> #info https://fedorahosted.org/rel-eng/report/10 the report for teh meeing items
16:38:33 <dgilmore> if you have anything you want discussed during the meeting add the meeting keyword
16:38:47 <dgilmore> tilldroid: okay. which ones?
16:40:06 <tilldroid> 5418
16:40:32 <tilldroid> Imho it should be closed wontfix
16:40:45 <dgilmore> tilldroid: thats really not a releng thing as we dont process the scm queue
16:41:10 <dgilmore> the processing of the queue is kind off a strange area though
16:41:21 <tilldroid> So it can be forwarded to fesco?
16:41:22 <dgilmore> ~there is a few people that have been doing it forever
16:41:46 <dgilmore> i honestly think we should just close it cantfix
16:41:58 <dgilmore> and there is nothing to do
16:42:33 <dgilmore> not sure why that ticket did not show up in the meeting report
16:44:26 <tyll_> another old ticket is https://fedorahosted.org/rel-eng/ticket/4084 - it is about multidep breakage
16:44:35 <tyll_> *multiarch
16:44:39 <dgilmore> i guess one option is to pull processing the SCM queue into releng
16:44:53 <dgilmore> but even then I wouldn't change it
16:45:49 <dgilmore> its sat there as without major work to mash we can not do anything
16:46:57 <dgilmore> I honestly do not know the best way to deal with it
16:47:58 <dgilmore> it boils down to mash would need to know whats multilibbed in relesase repos
16:48:11 <dgilmore> then make sure the same things get done in updates
16:48:27 <dgilmore> we would never be able to unmultilib something
16:49:19 <dgilmore> we would need to add a whole lot of heavyweight wrapping in the process to make sure of things
16:49:35 <dgilmore> the process is already heavy and time consuming
16:50:03 <dgilmore> there is no clear point where we could break things with history
16:51:36 <dgilmore> tyll_: not sure that answers things for you
16:52:20 <tyll_> dgilmore: maybe the bug can then get a more describtive title, because currently it sounds gcc specific, but it is actually some mash problem I do not really understand :-(
16:52:42 <tyll_> this might make it clear that it is nothing that can be done right now
16:52:47 <dgilmore> tyll_: yeah. and its not really a huge problem
16:53:21 <dgilmore> but it boils down to the way we do multilib in the repos depends on the contents of the repos
16:53:30 <dgilmore> the checking is self contained
16:54:22 <dgilmore> so if nothing in say the updates repo pulls in a package to be multilibbed but it was multilibbed in the release tree you may hit an issue
16:54:48 <rdieter> I'm of a mind to say wontfix (for now, at releng level at least)
16:55:12 <rdieter> and recommend focus on fixing the packaging bugs that cause the problem(s)
16:55:54 <dgilmore> rdieter: it may not be a packaging bug
16:56:18 <rdieter> I thought in this particular case it was, some -devel pkgs with a dep: gcc-gfortran(x86-32)
16:56:26 <dgilmore> foo-lib is pulled into being multillibed due to bar requiring it only
16:56:31 <dgilmore> foo is updated
16:56:50 * masta looks in
16:56:55 <masta> sry I'm late....
16:56:58 <dgilmore> but it doesnt get multilibbed in updates repo as bar wasnt rebuilt
16:57:29 <dgilmore> bar is multilibbed due to having a bar-devel sub-package
16:57:38 <dgilmore> foo doesnt for some reason
16:57:54 <dgilmore> honestly its a pulling at straws senario
16:58:19 <dgilmore> rdieter: yeah i think this case was a packaging bug
16:58:43 <tyll_> is there a mash specific bug tracker?
16:58:48 <dgilmore> tyll_: any other tickets you would like to discuss?
16:58:57 <dgilmore> tyll_: bugzilla
16:58:58 * rdieter doesn't follow that example exactly, but still would argue that until its understood fully to be *not* a packaging bug, the resources/energy required to fix/workaround at releng level simply isn't worth it
16:59:41 <tyll_> https://fedorahosted.org/rel-eng/ticket/3624
16:59:59 <tyll_> This sounds like it can be easily fixed or might already be fixed
17:00:30 <dgilmore> there is a script in puppet/ansible that gets run to update the filelists
17:01:47 <dgilmore> tyll_: so some issues
17:02:19 <dgilmore> there is 3 different machines that could be updating that fullfilelist file
17:02:30 <dgilmore> all could be doing it at once
17:02:40 <dgilmore> so really it needs some kind of network lock
17:02:48 <dgilmore> maybe using nfslock
17:03:08 <dgilmore> if the process cant get a lock it needs to wait
17:03:27 <dgilmore> as the process making the current file could miss the changes we are updating it for
17:04:40 <dgilmore> so its not fixed, it is fixable, the fix needs some thought to ensure processes dont fight each other
17:05:01 <dgilmore> tyll_: its the kind of thing id actually like to just pull out of the composedb
17:05:36 <tyll_> ah, ok, I'll update the ticket with this information
17:06:25 <dgilmore> http://paste.fedoraproject.org/66190/13890279/ is the current script thats used
17:06:56 <dgilmore> we could easily add compression
17:07:54 <nirik> it could also tie into fedmsg?
17:08:08 <nirik> (ie, something finished that would cause updates to the flielist)
17:08:19 <dgilmore> nirik: yeah
17:08:32 <dgilmore> so we could work a few different ways to redo the process
17:10:37 <dgilmore> anyway
17:10:56 <tyll_> I could take a look into fedmsg integration, but I do not have access to the system so it is pretty abstract to me what is happening there
17:11:48 <dgilmore> tyll_: you probably do, I can work with you later to make sure you do
17:12:43 <tyll_> dgilmore: ok
17:13:06 <dgilmore> okay lets move on
17:13:24 <dgilmore> #topic s390 status
17:13:37 <dgilmore> sharkcz: want to give us a quick status update
17:14:24 <sharkcz> I'm doing last steps before releasing, all stuff is in place, running hardlink right now
17:14:36 <dgilmore> okay cool
17:14:56 <sharkcz> I've rerun last branched with fixed inheritance and fixed key id so all mashed rpms are signed
17:15:10 <sharkcz> https://bugzilla.redhat.com/show_bug.cgi?id=1048189 is the update to mash configs
17:15:21 <dwa> yeah, the key ID thing drove me nuts too before figuring out what it was :(
17:15:55 <dgilmore> sorry i failed there
17:16:42 <dgilmore> sharkcz: hows updates etc?
17:17:06 <sharkcz> mostly done and signed, will mash them soon
17:17:18 <dgilmore> cool
17:17:32 <dgilmore> okay lets move on.
17:17:39 <dgilmore> #topic ppc status
17:17:44 <dgilmore> dwa: hows things ?
17:18:08 <dwa> We released F20 for Power before the break, and have been signing/mashing/pushing updates as normal
17:18:18 <dwa> this week we're planning on doing some koji maintenance - system updates, etc.
17:18:38 <dgilmore> cool
17:18:39 <dwa> we should also be receiving some new Power systems from IBM soon - they'll support both KVM virt (instead of PowerVM) and PPC64LE
17:18:47 <dgilmore> nice
17:18:53 <masta> cool
17:19:02 <dgilmore> so we will need to come up with a plan to do the ppc64le bootstrap
17:19:09 <dwa> last I heard we were waiting on loaner agreements etc. to get signed
17:19:32 <dwa> yeah, once we get hardware we'll be able to actually do stuff. last I heard from IBM the ppc64le ABI was still a little unstable, though
17:20:05 <dgilmore> :(
17:20:14 <dwa> so they were doing more rebuilds from scratch internally
17:20:35 <dwa> but hopefully they'll find all the bugs and everything will work perfectly the first time for us ;)
17:21:19 <dgilmore> that would be nice
17:22:05 <dgilmore> okay
17:22:13 <dwa> that's it from me
17:22:17 <dgilmore> #topic arm status
17:22:28 <dgilmore> aarch64 has has a few bugs to get fixed low down
17:22:34 <dgilmore> but is slowly getting there
17:22:51 <dgilmore> going to look at enabling rawhide this week
17:23:02 <dgilmore> 32 bit updates slowly are gtting out
17:23:26 <dgilmore> with f18 eol next week softfp will be dead
17:23:27 <masta> dgilmore: what low down things in aarch64?
17:24:20 <masta> dgilmore: stuff related to using the foundation?
17:24:49 <dgilmore> masta: some bugs in elfutils i believe
17:25:33 <dgilmore> okay
17:25:38 <dgilmore> #topic open floor
17:25:46 <dgilmore> anyone have anything they want to discuss
17:26:12 <masta> I did the updates push last two weeks, somebody else turn
17:26:45 <masta> we should probably setup a fedocal for this
17:26:45 <dgilmore> dwa: want to?
17:26:56 <dgilmore> masta: yeah thats on the list to do
17:27:29 <masta> I think dwa was last week, but pto... so I covered
17:27:30 <dwa> yeah, happy to
17:27:46 <dwa> and agree, a calendar somewhere would be excellent :)
17:27:59 <dwa> masta: thanks for that, btw
17:28:18 <dgilmore> okay dwa  for updates this week
17:28:41 <tyll_> I would like to propose to somehow organise the trac items so that task-like items can be better separated from other items that only need to be done eventually such as tooling changes
17:29:07 <dgilmore> tyll_: okay, i can make you an admin to make catagories etc
17:29:13 <dgilmore> im going to fossdem and devconf at the end of the month
17:29:36 <dgilmore> as devconf im giving a talk on future things for releng and how people can help to deliver them
17:30:21 <tyll_> dgilmore: I am planning to go to fosdem as well
17:31:03 <dgilmore> tyll_: excellent will be good to meet you finally
17:31:48 <masta> =)
17:33:41 <dgilmore> over the christmas break i worked on changing my blog
17:33:54 <dgilmore> im planning on writting more often about whats going on in releng
17:34:12 <dgilmore> and try spark interest in people to get involved
17:34:21 <dgilmore> anything else?
17:34:29 <dgilmore> if not we just have gon over an hour
17:34:35 <dgilmore> ill wrap up
17:35:27 <dgilmore> #endmeeting