ReleaseEngineering/Meetings/2007-may-07

Fedora Release Engineering Meeting :: Monday 07-MAY-07

Status of the Merge

• f13
• I'm poking at some software bill wrote a while ago to build package repos out of Koji.
• I'm somewhat close to having a repo of packages created in a test environment.
• expected to be at this point on Thursday, but well, we all forgot about ppc64
• some of the extras importing didn't go as quickly as I 'd hoped.
• plan for now I _think_ it will be koji -> repo of packages as rawhide
• people can use pungi to make installable trees, I'll probably make one somewhere.
• I need to work on the multilib stuff to make sure it's multilibbing like distill did.
• no plan yet around needsign/package signing
• ppc64 things are still uploading/importing; build status is unknown
• found a few things that didn't have matching NVRs did not build--libupnp-1.4.3-1.fc7 for example
• warren
• volunteers to be "bit mover" while f13 is at the Red Hat Summit
• jwb
• need a document explaining how to get your package into the release... something like "If you did FOO before the merge, here is how to do FOO now"
• will take a crack at the before/after document
• lengthy technical discussion around using auto-add script and building packages with koji
• all builds must be performed manually
• no specific decisions were made
• notting will work on getting automated builds working over the next three days
• notting
• there's some stuff that won't build due to various errors - i've filed some bugs, but still have more to file
• ocaml and maven are being worked on
• things that BR mono will need ifarched
• i'll try and attack maven again later *ugh*
• everything on the missing dep tracker should be OK now
• 1224 SRPMS built so far, give or take a few
• using ibiblio--mmcgrath and f13 will discuss at Summit

mirror layout and getting bits to said mirror

• f13
• Since we're still doing Extras for 6/5, perhaps we just tie into that rsync process to get development synced as well? Compse rawhide to the place that is servied by buildsys: rsync?
• symlinks or hardlinks are the plan.
• nirik: proposed mirror layout changes look sane
• No formal decisions made

Building Rawhide

• not going to have a post-fc7 rawhide until fc7 goes gold
• important to be able to use rawhide to tackle F7 blocker BZs
• rawhide push will be good for checking EVR and broken deps issues.
• can't make rawhide automated today; will try to enable eventually. No date set.
• nirik: should we eventually revisit signing rawhide?
• would possibly make it so yum could detect what repo a package was from and help with all the repotag nightmares?
• f13:
• I'd like rawhide to be signed, with the buildsystem throwaway key but that'll involve an automated signing system.
• we have a signing system; manual or automated
• discussion of getting a "real" rawhide working before we branch and also how/when to branch
• no decision reached

F7 Blocker List

• https://bugzilla.redhat.com/bugzilla/showdependencytree.cgi?id=150226&hide_resolved=1
• 104 open bugs
• 17 Days until release
• wwoods
• will do some blocker bug triage
• will loudly announce
• some bugs might not be getting addressed because developers are waiting for builds of rawhide

Schedule Slip

• f13
• not entirely convinced we'll be ready for deep freeze by Thursday, though probably 'frozen enough to branch cvs'
• okay with more slipping because what we just did (merge) was _huge_ and we're not coming back online as fast as I hoped we would.
• can't feel good about deep freezing only a couple days after a merged repo shows up.
• decision reached:
• slip 1 week on the freeze and re-evaluate blockers next week
• no explicit decision made on GA date
• state that merger took longer/more effort than panned for
• need more time for rawhide to settle down a bit and maintianers to get their builds in.

Build System Outage

• outage on Wed for about an hour and a half. We're getting some new filers installed.
• Time: TBD

IRC Transcript

f13	ok, who's here.  Role call.	<a href="#t13:05" class="time">13:05</a>
* rdieter here		<a href="#t13:05" class="time">13:05</a>
f13 changed the topic of #fedora-meeting to: Fedora Release Team Meeting		<a href="#t13:05" class="time">13:05</a>
* wwoods here		<a href="#t13:05" class="time">13:05</a>
* mclasen is probably going to drop out at some point		<a href="#t13:06" class="time">13:06</a>
* mmcgrath here		<a href="#t13:06" class="time">13:06</a>
* warren here		<a href="#t13:07" class="time">13:07</a>
poelcat	present	<a href="#t13:07" class="time">13:07</a>
jwb	here	<a href="#t13:07" class="time">13:07</a>
* jeremy is here		<a href="#t13:07" class="time">13:07</a>
jeremy	(mostly)	<a href="#t13:07" class="time">13:07</a>
warren	Do we need davej?	<a href="#t13:07" class="time">13:07</a>
jwb	when do we not need davej?	<a href="#t13:07" class="time">13:07</a>
jeremy	he may still be in nyc	<a href="#t13:07" class="time">13:07</a>
f13	heh	<a href="#t13:08" class="time">13:08</a>
f13	we could use a notting.	<a href="#t13:08" class="time">13:08</a>
f13	I"m txting him.	<a href="#t13:08" class="time">13:08</a>
rdieter	f13: the previous summoning failed? :)	<a href="#t13:08" class="time">13:08</a>
warren	crazy kids	<a href="#t13:08" class="time">13:08</a>
f13	rdieter: indeed	<a href="#t13:08" class="time">13:08</a>
warren	rdieter, if we summon now, we'll be stuck for 5 minutes in the summoning animation.	<a href="#t13:08" class="time">13:08</a>
jwb	"r u thr"  "ur hot.  need u now"	<a href="#t13:09" class="time">13:09</a>
f13	jwb: creepy.  it's like you're watching my phone over my shoulder.	<a href="#t13:10" class="time">13:10</a>
jwb	koji has altered my state of mind and granted me new found powers	<a href="#t13:10" class="time">13:10</a>
rdieter	all bow down to the almighty koji-bot.	<a href="#t13:11" class="time">13:11</a>
wwoods	heh jwb is a mighty build-taggin' machine	<a href="#t13:11" class="time">13:11</a>
warren	* jwb is now known as koji-bot.	<a href="#t13:14" class="time">13:14</a>
jwb	phear me	<a href="#t13:14" class="time">13:14</a>
wwoods	so how far are we from having koji-rawhide?	<a href="#t13:15" class="time">13:15</a>
f13	hrm, no notting, so whatever.	<a href="#t13:15" class="time">13:15</a>
f13	Status of the merge.	<a href="#t13:15" class="time">13:15</a>
f13	wwoods: I'm poking at some software bill wrote a while ago to build package repos out of Koji.  I'm somewhat close to having a repo of packages created in a test environment.	<a href="#t13:16" class="time">13:16</a>
jwb	the "natives formerly known as extras maintainers" are getting restless	<a href="#t13:16" class="time">13:16</a>
mmcgrath	brb	<a href="#t13:16" class="time">13:16</a>
f13	yeah, tell me about it	<a href="#t13:16" class="time">13:16</a>
f13	I expected to be at this point on Thursday, but well, we all forgot about ppc64	<a href="#t13:16" class="time">13:16</a>
f13	and some of the extras importing didn't go as quickly as I 'd hoped.	<a href="#t13:17" class="time">13:17</a>
wwoods	so is the plan to do koji--pungi-->rawhide all on the external machines, and everyone will mirror from there?	<a href="#t13:17" class="time">13:17</a>
warren	With people gone at the summit, I *hope* we can get at least rawhide trees going before they leave.  we would be in MUCH better shape.	<a href="#t13:17" class="time">13:17</a>
f13	wwoods: actually the plan for now I _think_ it will be koji -> repo of packages as rawhide	<a href="#t13:17" class="time">13:17</a>
warren	I'm going to be stuck playing "keep developers happy" while things are broken this week.	<a href="#t13:17" class="time">13:17</a>
wwoods	ah, right on	<a href="#t13:17" class="time">13:17</a>
f13	and then people can use pungi to make installable trees, I'll probably make one somewhere.	<a href="#t13:17" class="time">13:17</a>
f13	warren: given that I'm gone at the summit too, I hope that as well.	<a href="#t13:18" class="time">13:18</a>
wwoods	so what are the roadblocks for making the repo from koji? I'm happy to help	<a href="#t13:18" class="time">13:18</a>
f13	I may spend many a summit night doing stuff with it though.	<a href="#t13:18" class="time">13:18</a>
f13	wwoods: really it's just tweaking the code.	<a href="#t13:18" class="time">13:18</a>
f13	I need to work on the multilib stuff I think, to make sure it's multilibbing like distill did.	<a href="#t13:18" class="time">13:18</a>
warren	are we settled on the new tree structure for the mirrors?  There was a question about updates-testing, was that resolved?	<a href="#t13:19" class="time">13:19</a>
f13	and we need some what to weed out noarch packages that aren't supposed to be on said arch (have to trace back to srpm and look at ExcludeArch/ExclusiveArch in the srpm)	<a href="#t13:19" class="time">13:19</a>
f13	warren: what was the question?	<a href="#t13:19" class="time">13:19</a>
warren	f13, somebody's mail indicated that updates-testing was missing from the new tree sturcture	<a href="#t13:19" class="time">13:19</a>
* warren finds the mail		<a href="#t13:19" class="time">13:19</a>
f13	If somebody wants to play bit-mover and move some bits around for mirror staging that would be kind of cool.	<a href="#t13:19" class="time">13:19</a>
f13	<a href="http://fedoraproject.org/wiki/Infrastructure/CoreExtrasMerge#head-d370223ce90e602f6d5c4bf3584bd20800da68b6">http://fedoraproject.org/wiki/Infrastructure/CoreExtrasMerge#head-d370223ce90e602f6d5c4bf3584bd20800da68b6</a>  I see testing there.	<a href="#t13:20" class="time">13:20</a>
warren	f13, ok, I dno't know what he was talking about.  nm	<a href="#t13:20" class="time">13:20</a>
warren	f13, I can play bitmover, but I need more context of what needs to be done and where.	<a href="#t13:20" class="time">13:20</a>
warren	i'll be here all week	<a href="#t13:20" class="time">13:20</a>
warren	People will blame me for things being broken in both core and extras =)	<a href="#t13:21" class="time">13:21</a>
* wwoods is also not going to the summit		<a href="#t13:21" class="time">13:21</a>
warren	ok good wwoods can share the blame.	<a href="#t13:21" class="time">13:21</a>
wwoods	totally	<a href="#t13:21" class="time">13:21</a>
f13	notting is on the way.	<a href="#t13:21" class="time">13:21</a>
* warren waits five minutes.		<a href="#t13:22" class="time">13:22</a>
f13	warren: unfortunately bit moving is rather involved :(	<a href="#t13:22" class="time">13:22</a>
warren	f13, I want to know what is involved	<a href="#t13:23" class="time">13:23</a>
f13	warren: to be perfectly honest I'm not entirely sure which part needs to be done where.  I drafted the layout, but didn't necessarily write down where each move has to happen.	<a href="#t13:23" class="time">13:23</a>
wwoods	looking over the tasklist - what's the plan for needsign / package signing?	<a href="#t13:23" class="time">13:23</a>
f13	wwoods: open question.	<a href="#t13:24" class="time">13:24</a>
jwb	whilst we wait... is someone working on a "Here's wtf you need to do with koji" doc for former-Extras package maintainers?	<a href="#t13:24" class="time">13:24</a>
f13	wwoods: Koji has the ability to keep a signed header in the database.  I think I can make some code magically appear that deals with copying packages locally, signing them with gpg, and importing the signed header into Koji, potentially writing out the signed package to the file system.	<a href="#t13:24" class="time">13:24</a>
f13	wwoods: that code will probably need some tuning though.	<a href="#t13:25" class="time">13:25</a>
wwoods	jwb: /usr/share/doc/koji-1.1/docs/HOWTO.html seems pretty dang good so far	<a href="#t13:25" class="time">13:25</a>
jwb	wwoods, that's not what i mean	<a href="#t13:25" class="time">13:25</a>
wwoods	jwb: ah, okay	<a href="#t13:25" class="time">13:25</a>
jwb	wwoods, i meant "this is wtf you need to do to get your packages into $release"	<a href="#t13:26" class="time">13:26</a>
f13	wwoods: I think he's looking for more of the "If you did FOO before the merge, here is how to do FOO now"	<a href="#t13:26" class="time">13:26</a>
jwb	right, what f13 said	<a href="#t13:26" class="time">13:26</a>
wwoods	right right	<a href="#t13:26" class="time">13:26</a>
notting	f13: sorry	<a href="#t13:26" class="time">13:26</a>
wwoods	yeah, that's important stuff. sadly I know very little about how Extras worked pre-merge.	<a href="#t13:26" class="time">13:26</a>
jwb	wwoods, people built packages and they were pushed	<a href="#t13:27" class="time">13:27</a>
warren	One big issue that came up is adding of new packages... I was misinformed earlier that auto-populate from owners.list to koji would happen.	<a href="#t13:27" class="time">13:27</a>
warren	I need to know exactly howt to add packages to koji.  I might be able to tie it into the CVS adding procedure.	<a href="#t13:27" class="time">13:27</a>
jwb	wwoods, there's lots of stuff now that they have to deal with.  rel-eng requests during freeze, updates stuff, etc	<a href="#t13:27" class="time">13:27</a>
jwb	warren, f13 sent us an email on that	<a href="#t13:27" class="time">13:27</a>
warren	jwb, where?	<a href="#t13:27" class="time">13:27</a>
warren	jwb, I've been swamped	<a href="#t13:28" class="time">13:28</a>
notting	warren: we have a script that can auto-add, it's just not on	<a href="#t13:28" class="time">13:28</a>
warren	notting, ok, so I need not worry about this?	<a href="#t13:28" class="time">13:28</a>
warren	notting, you going to summit too?	<a href="#t13:28" class="time">13:28</a>
warren	notting, possible to turn it on before summit?	<a href="#t13:28" class="time">13:28</a>
jwb	warren, to cvsadmin-members	<a href="#t13:28" class="time">13:28</a>
jwb	warren, in short: koji add-pkg --owner <FAS user id> dist-fc7 <package>	<a href="#t13:29" class="time">13:29</a>
warren	jwb, although with notting's mention of auto-add, do we need to do this or not?	<a href="#t13:29" class="time">13:29</a>
notting	warren: well, it will still be a delay for whatever to run out of cron	<a href="#t13:29" class="time">13:29</a>
warren	notting, how often?  can it run as often as the acl script?	<a href="#t13:29" class="time">13:29</a>
warren	that would be VERY logical	<a href="#t13:29" class="time">13:29</a>
jwb	or be part of the acl script?	<a href="#t13:29" class="time">13:29</a>
f13	I just realized though	<a href="#t13:29" class="time">13:29</a>
f13	adding the package to dist-fc7 isn't enough, it has to be added to f7-final too	<a href="#t13:30" class="time">13:30</a>
warren	f13, we can adjust the script as needed	<a href="#t13:30" class="time">13:30</a>
f13	f7-final doesn't inherit from dist-fc7 (as then builds would be overridden)	<a href="#t13:30" class="time">13:30</a>
f13	warren: nod, it just has to be added twice	<a href="#t13:30" class="time">13:30</a>
f13	although.	<a href="#t13:30" class="time">13:30</a>
jwb	f13, what?	<a href="#t13:30" class="time">13:30</a>
warren	notting, how often can we run auto-add?  why not do it as part of the 30 minute acl gen.	<a href="#t13:30" class="time">13:30</a>
f13	I suppose we could make inheritance work.  hrm.	<a href="#t13:30" class="time">13:30</a>
jwb	f13, why are we adding new packages to f7-final?	<a href="#t13:30" class="time">13:30</a>
* jwb is slightly confused		<a href="#t13:30" class="time">13:30</a>
f13	I am too.	<a href="#t13:31" class="time">13:31</a>
f13	I don't remember what we decided about new packages being added to hte collection during this time	<a href="#t13:31" class="time">13:31</a>
warren	jwb, people continue to add things to Extras now.  We want packages in FE6 to be newer than F7?	<a href="#t13:31" class="time">13:31</a>
f13	whether they'd land in rawhide or be in the Everything tree	<a href="#t13:31" class="time">13:31</a>
warren	oh!	<a href="#t13:31" class="time">13:31</a>
warren	How about this:	<a href="#t13:31" class="time">13:31</a>
notting	warren: well, it would require me putting my koji certs somewhere i may not want to. aside from that...	<a href="#t13:31" class="time">13:31</a>
warren	auto-add to dist-fc7, allow builds there.	<a href="#t13:31" class="time">13:31</a>
warren	but only allow in f7-final if requested and approved by rel-eng.	<a href="#t13:32" class="time">13:32</a>
warren	which might be rare	<a href="#t13:32" class="time">13:32</a>
notting	f13: so we have policy that there will be stuff in the buildroot we won't ship?	<a href="#t13:32" class="time">13:32</a>
jwb	that's what i've been doing	<a href="#t13:32" class="time">13:32</a>
* notting is confused		<a href="#t13:32" class="time">13:32</a>
warren	notting, where is acl gen currently running?	<a href="#t13:32" class="time">13:32</a>
notting	cvs	<a href="#t13:32" class="time">13:32</a>
f13	notting: unfortunately not.  This is the really sucky sucky part of releasing a distro	<a href="#t13:32" class="time">13:32</a>
warren	notting, generate a non-notting user with the koji perm?	<a href="#t13:32" class="time">13:32</a>
f13	we basically have two options.	<a href="#t13:33" class="time">13:33</a>
f13	option A) no builds get into build roots without rel-eng interaction.	<a href="#t13:33" class="time">13:33</a>
notting	warren: don't think a non-person admin koji user is *better*	<a href="#t13:33" class="time">13:33</a>
f13	keeps buildroots sane.  Adds a _ton_ of lag gime when somebody needs to build an interdependent stack of packages.	<a href="#t13:33" class="time">13:33</a>
warren	notting, in the long-term what option do we have?  If we cannot trust cvs-int, aren't we screwed?	<a href="#t13:33" class="time">13:33</a>
warren	notting, we have to automate this *somehow* or we lose.	<a href="#t13:33" class="time">13:33</a>
f13	option B) buildroots continue to self populate, need rel-eng to interact to get your stack into rawhide/final.	<a href="#t13:33" class="time">13:33</a>
notting	warren: why should koji necessarily trust it over other machines?	<a href="#t13:34" class="time">13:34</a>
warren	Option B is OK if we have nagmail and pull in packages that were part of the buildroot when we tag a f7-final package.	<a href="#t13:34" class="time">13:34</a>
notting	warren: i'm not saying i can't put the perms there	<a href="#t13:34" class="time">13:34</a>
f13	with option B it is far easier to prepare a stack of packages for release, but then you have buildroots with stuff that isn't necessarily shipped.	<a href="#t13:34" class="time">13:34</a>
warren	notting, 1) We need automation or we lose.  2) cvs is supposed to be one of our most secure boxes, we're screwed if it isn't.  3) why not just do it?	<a href="#t13:34" class="time">13:34</a>
notting	warren: *shrug* i'm not going to say we 'lose' if we have to run one simple command-line program, but, eh, whatever	<a href="#t13:35" class="time">13:35</a>
notting	warren: more relevant: cvs runs rhel4. koji?	<a href="#t13:36" class="time">13:36</a>
warren	notting, what other option we have?  we really can't handle adding of new packages manually.	<a href="#t13:36" class="time">13:36</a>
notting	warren: a) we do now b) offtopic?	<a href="#t13:36" class="time">13:36</a>
warren	notting, I mean, we already do it manually for cvs, it only makes sense that the fruit of that effort go into koji, it is the same problem.	<a href="#t13:36" class="time">13:36</a>
f13	notting: koji client works on RHEL4	<a href="#t13:37" class="time">13:37</a>
warren	notting, *I* am being blamed for packages not being buildable in koji.  This should be automated especially if we have a script to do so.	<a href="#t13:37" class="time">13:37</a>
* f13 wonders if we can get back on topic a bit.		<a href="#t13:38" class="time">13:38</a>
warren	This is totally on topic.	<a href="#t13:38" class="time">13:38</a>
warren	We're stuck on this.	<a href="#t13:38" class="time">13:38</a>
warren	We need this to keep the contributors happy.	<a href="#t13:38" class="time">13:38</a>
f13	warren: basically the response I have right now is "tough".  Things are going to be rough.  We may have to do many things by hand right now, deal.	<a href="#t13:39" class="time">13:39</a>
warren	notting, where is the auto-add script?  I'll use it myself for now.	<a href="#t13:39" class="time">13:39</a>
notting	warren: in the same place as everything else	<a href="#t13:40" class="time">13:40</a>
notting	and, no offense, but mail was sent as to how to add packages by hand - if you're adding packages, you need to read that	<a href="#t13:40" class="time">13:40</a>
warren	I misunderstood that it was happening automatically	<a href="#t13:41" class="time">13:41</a>
warren	I'll deal with it manually for now.	<a href="#t13:41" class="time">13:41</a>
warren	ok, move on.	<a href="#t13:41" class="time">13:41</a>
notting	warren: i'll get it set up automatically when i can. might not be for a day or three	<a href="#t13:41" class="time">13:41</a>
warren	notting, cool, thanks	<a href="#t13:42" class="time">13:42</a>
f13	notting: how far along are we in the ppc64 stuff?	<a href="#t13:42" class="time">13:42</a>
notting	f13: fairly far build-wise. dunno about import	<a href="#t13:42" class="time">13:42</a>
* notting argues with koji urls		<a href="#t13:42" class="time">13:42</a>
notting	f13: figure out what was up with docbook2X?	<a href="#t13:44" class="time">13:44</a>
f13	notting: I sent mail to you.	<a href="#t13:44" class="time">13:44</a>
notting	ok	<a href="#t13:44" class="time">13:44</a>
f13	notting: if you don't have your certs and koji config in place, querying for stuff won't work too well.	<a href="#t13:45" class="time">13:45</a>
notting	f13: yeah, but afaik they were. oh well, it's there now	<a href="#t13:45" class="time">13:45</a>
notting	as for ppc64	<a href="#t13:45" class="time">13:45</a>
f13	things are still uploading/importing	<a href="#t13:45" class="time">13:45</a>
warren	notting, is the acl gen script still running every 30 minutes?  (just not bugzilla sync?)	<a href="#t13:45" class="time">13:45</a>
notting	warren: yep	<a href="#t13:45" class="time">13:45</a>
f13	I started the script over again looking at /local/	<a href="#t13:45" class="time">13:45</a>
warren	is bugzilla sync coming back?	<a href="#t13:45" class="time">13:45</a>
notting	f13: there's another directory, i'll start it over that later. but that extra stuff is stuff that shouldn't be build-reqs	<a href="#t13:46" class="time">13:46</a>
notting	once everything that's built now is imported, we should be *reasonably* good to go for ppc64	<a href="#t13:46" class="time">13:46</a>
f13	notting: ok.  I found a few things that didn't have matching NVRs	<a href="#t13:46" class="time">13:46</a>
notting	there's some stuff that won't build due to various errors - i've filed some bugs	<a href="#t13:46" class="time">13:46</a>
notting	ocaml is being worked on	<a href="#t13:46" class="time">13:46</a>
f13	libupnp-1.4.3-1.fc7 for example	<a href="#t13:46" class="time">13:46</a>
notting	things that BR mono will need ifarched	<a href="#t13:46" class="time">13:46</a>
notting	and i'll try and attack maven again later *ugh*	<a href="#t13:47" class="time">13:47</a>
f13	looks like libupnp was rebuilt later IN koji and successfully pooped out a ppc64 package	<a href="#t13:47" class="time">13:47</a>
notting	f13: good.	<a href="#t13:48" class="time">13:48</a>
notting	everything on the missing dep tracker should be OK now	<a href="#t13:48" class="time">13:48</a>
f13	ok.	<a href="#t13:51" class="time">13:51</a>
f13	notting: so then it's just maven, ocaml, and the things you've filed bugs on?	<a href="#t13:52" class="time">13:52</a>
f13	(left to build)	<a href="#t13:52" class="time">13:52</a>
notting	f13: more than that, as i haven't filed bugs on everything	<a href="#t13:53" class="time">13:53</a>
f13	ok.	<a href="#t13:53" class="time">13:53</a>
* notting idly wonders how many of the build failures are things that don't build on any arch ATM :)		<a href="#t13:54" class="time">13:54</a>
notting	f13: 1224 SRPMS built, give or take a few	<a href="#t13:55" class="time">13:55</a>
f13	that's not a small number.	<a href="#t13:56" class="time">13:56</a>
f13	so the next thing to talk about.	<a href="#t13:56" class="time">13:56</a>
f13	mirror layout and getting bits to said mirror.	<a href="#t13:57" class="time">13:57</a>
mmcgrath	f13: can we start using ibiblio soon?	<a href="#t13:57" class="time">13:57</a>
f13	Since we're still doing Extras for 6/5, perhaps we just tie into that rsync process to get development synced as well?  Compse rawhide to the place that is servied by buildsys: rsync?	<a href="#t13:57" class="time">13:57</a>
f13	mmcgrath: you'll be at summit right?  Can we talk there about it ?  I just can't handle more changes.	<a href="#t13:58" class="time">13:58</a>
mmcgrath	f13: this shouldn't involve any changes on your side if things.  Lets talk at the summit.	<a href="#t13:58" class="time">13:58</a>
* nirik wonders if it's possible to hardlink so mirrors don't get piles of bits all at once, and then remove the old dirs later?		<a href="#t13:58" class="time">13:58</a>
mmcgrath	s/if/of/	<a href="#t13:58" class="time">13:58</a>
f13	nirik: I do believe symlinks or hardlinks are the plan.	<a href="#t13:58" class="time">13:58</a>
f13	nirik: have you seen the proposed mirror layout changes?	<a href="#t13:59" class="time">13:59</a>
nirik	cool. Re-syncing all of devel core/extras would be anoying.	<a href="#t13:59" class="time">13:59</a>
nirik	yeap. Looks sane to me.	<a href="#t13:59" class="time">13:59</a>
notting	except, with content of /extras moving to a different toplevel, hardlinks may not help depending on how you have your sync set up. but, *shrug*	<a href="#t13:59" class="time">13:59</a>
f13	yeah, this isn't going to be fun.	<a href="#t14:00" class="time">14:00</a>
notting	oof	<a href="#t14:00" class="time">14:00</a>
notting	well, getting bits out:	<a href="#t14:01" class="time">14:01</a>
notting	- updates and releases are still on-demand, right?	<a href="#t14:01" class="time">14:01</a>
notting	so, the question is, do we make rawhide automated or not	<a href="#t14:02" class="time">14:02</a>
f13	notting: I don't think I can make rawhide automated today	<a href="#t14:02" class="time">14:02</a>
f13	eventually, yes I"d like it to be automated.	<a href="#t14:03" class="time">14:03</a>
notting	f13: step 1. get trees. then worry about push :)_	<a href="#t14:03" class="time">14:03</a>
f13	heh	<a href="#t14:03" class="time">14:03</a>
f13	notting: re the ppc64 stuff, /local/ is static now?  No more builds landing there?	<a href="#t14:03" class="time">14:03</a>
f13	when was the last build that landed there?	<a href="#t14:03" class="time">14:03</a>
notting	f13: more stuff landing there shortly	<a href="#t14:04" class="time">14:04</a>
notting	that's the 'needs to be available for building against' destination	<a href="#t14:04" class="time">14:04</a>
nirik	perhaps we could even revisit signing rawhide? :) (would possibly make it so yum could detect what repo a package was from and help with all the repotag nightmares)	<a href="#t14:04" class="time">14:04</a>
notting	and since matthias just built something in the build chain, i can build a few more things now	<a href="#t14:04" class="time">14:04</a>
f13	notting: argle.  Ok.	<a href="#t14:05" class="time">14:05</a>
f13	nirik: yes, eventually I'd like rawhide to be signed, with the buildsystem throwaway key	<a href="#t14:05" class="time">14:05</a>
f13	nirik: but that'll involve an automated signing system.	<a href="#t14:05" class="time">14:05</a>
f13	and we don't even have a signing system.	<a href="#t14:05" class="time">14:05</a>
f13	manual or automated.	<a href="#t14:05" class="time">14:05</a>
nirik	yep. Not something to worry about now, just somethign to keep in mind for down the road.	<a href="#t14:06" class="time">14:06</a>
mmcgrath	does the pkgdb have anything like that built into it?	<a href="#t14:06" class="time">14:06</a>
wwoods	I'd really like it if rawhide packages were being signed by the time F8 gets close	<a href="#t14:07" class="time">14:07</a>
* notting points wwoods at the pony		<a href="#t14:07" class="time">14:07</a>
wwoods	yes, that too! also a flamethrower.	<a href="#t14:07" class="time">14:07</a>
nirik	is the devel branching in cvs still planned for thursday? or is everyone going to be summiting?	<a href="#t14:08" class="time">14:08</a>
f13	not everybody	<a href="#t14:08" class="time">14:08</a>
f13	nirik: I'm not entirely convinced we'll be ready for deep freeze by Thursday	<a href="#t14:09" class="time">14:09</a>
f13	wwoods has been grumbling about the blocker list	<a href="#t14:09" class="time">14:09</a>
notting	f13: 'frozen enough to branch cvs'	<a href="#t14:09" class="time">14:09</a>
f13	well yeah, I suppose.	<a href="#t14:09" class="time">14:09</a>
warren	wwoods, +1, although we really don't need to worry about that for now.	<a href="#t14:10" class="time">14:10</a>
nirik	perhaps a plea to fedora-devel/maintainers/qa to have people poke at blocker bugs might be worth while?	<a href="#t14:10" class="time">14:10</a>
mmcgrath	FYI guys, there's going to be a build system outage on Wed for about an hour and a half.  (They're installing some new filers for us)	<a href="#t14:10" class="time">14:10</a>
mmcgrath	I'll send a note to the list when I get more details about the exact time.	<a href="#t14:10" class="time">14:10</a>
warren	A plea to developers about the blocker list will do us little good if we don't have rawhide trees for them to test.	<a href="#t14:11" class="time">14:11</a>
wwoods	Yeah the blocker list has 104 open bugs on it	<a href="#t14:11" class="time">14:11</a>
wwoods	and we have.. 17 days until release?	<a href="#t14:11" class="time">14:11</a>
* warren is upset at a core developer who refuses to fix a blocker because "rawhide isn't even pushing, I'll fix it later."		<a href="#t14:11" class="time">14:11</a>
f13	mmcgrath: oh thank mother of heaven.	<a href="#t14:11" class="time">14:11</a>
wwoods	warren: who the hell said that? half the beauty of koji is that people can pull packages straight from the buildsystem now	<a href="#t14:12" class="time">14:12</a>
warren	wwoods, caillon	<a href="#t14:12" class="time">14:12</a>
mmcgrath	f13: I'll believe it when I mount it :)	<a href="#t14:12" class="time">14:12</a>
wwoods	rawhide is far less essential now than it was	<a href="#t14:12" class="time">14:12</a>
warren	wwoods, it was just an excuse.	<a href="#t14:12" class="time">14:12</a>
f13	mmcgrath: dirty	<a href="#t14:12" class="time">14:12</a>
warren	wwoods, it is a valid argument that you can't upgrade to hte latest packages easily.  his argument is that people wont even see his changes.	<a href="#t14:12" class="time">14:12</a>
f13	Ok, anything hot on anybody's mind yet, or can I go back to trying to mash things together?	<a href="#t14:12" class="time">14:12</a>
wwoods	yeah I don't wanna be all chicken little here but seriously, 3 weeks! 104 bugs!	<a href="#t14:12" class="time">14:12</a>
warren	+1 to mash!	<a href="#t14:13" class="time">14:13</a>
notting	wwoods: wheee	<a href="#t14:13" class="time">14:13</a>
f13	wwoods: nod.  I'm OK with more slipping.  What we just did was _huge_ and we're not coming back online as fast as I hoped we would.	<a href="#t14:13" class="time">14:13</a>
warren	wwoods, go ahead and announce loudly the blocker list, but I think rawhide pushing is going to be very important for the blocker list to be tackled.	<a href="#t14:13" class="time">14:13</a>
wwoods	warren: agreed	<a href="#t14:13" class="time">14:13</a>
jwb	wwoods, koji rocks but it is not a replacement for a repo yum can pull from	<a href="#t14:13" class="time">14:13</a>
wwoods	yeah probably I should do some blocker triage	<a href="#t14:13" class="time">14:13</a>
nirik	also rawhide push will be good for checking EVR and broken deps issues.	<a href="#t14:14" class="time">14:14</a>
f13	jwb: I think I can get a repo to pull from very quickly	<a href="#t14:14" class="time">14:14</a>
jwb	f13, would be awesome	<a href="#t14:14" class="time">14:14</a>
wwoods	jwb: yeah, but the absence of rawhide is no longer a blocker to getting packages in people's hands to test fixes	<a href="#t14:14" class="time">14:14</a>
f13	it'll just be non-multilib and not mirrored.	<a href="#t14:14" class="time">14:14</a>
* nirik suspects there will be lots of EVR problems. ;(		<a href="#t14:14" class="time">14:14</a>
f13	and having tonns of people pull from it is going to not be good for the current filer we're using.	<a href="#t14:14" class="time">14:14</a>
jwb	wwoods, individuals, no.  testers as a collective, yes	<a href="#t14:14" class="time">14:14</a>
wwoods	jwb: right	<a href="#t14:14" class="time">14:14</a>
f13	jwb: so I'd rather wait until we have _real_ filers and maybe some other box that can http serve it instead of the koji server itself.	<a href="#t14:14" class="time">14:14</a>
wwoods	since we're talkin' about the blocker list here, I'm thinking about individuals	<a href="#t14:14" class="time">14:14</a>
wwoods	(i.e. bug reporters)	<a href="#t14:15" class="time">14:15</a>
f13	absolutely	<a href="#t14:15" class="time">14:15</a>
jwb	ok	<a href="#t14:15" class="time">14:15</a>
mclasen	wwoods: those 104 bugs on the blocker list are self-nominated blockers though, right ?	<a href="#t14:15" class="time">14:15</a>
wwoods	rawhide is obviously hugely important, but it's not The Only Way Anyone Can Test Anything anymore	<a href="#t14:15" class="time">14:15</a>
wwoods	mclasen: no, those are put there by, basically, the people in this meeting	<a href="#t14:15" class="time">14:15</a>
f13	wwoods: oh the hidden blocker?	<a href="#t14:16" class="time">14:16</a>
wwoods	although people *can* add their own, we generally pull them off if they're silly	<a href="#t14:16" class="time">14:16</a>
wwoods	so for the most part the blocker list is for-real	<a href="#t14:16" class="time">14:16</a>
mclasen	wwoods: well, if the blocker list is secret, how are people supposed to know that the release is blocking on them ?	<a href="#t14:16" class="time">14:16</a>
wwoods	it's not secret	<a href="#t14:16" class="time">14:16</a>
mclasen	which blocker bug are we talking about ?	<a href="#t14:16" class="time">14:16</a>
f13	my misword.	<a href="#t14:16" class="time">14:16</a>
wwoods	I think f13 was kidding?	<a href="#t14:16" class="time">14:16</a>
wwoods	FC7Blocker	<a href="#t14:17" class="time">14:17</a>
wwoods	<a href="https://bugzilla.redhat.com/bugzilla/showdependencytree.cgi?id=150226&hide_resolved=1">https://bugzilla.redhat.com/bugzilla/showdependencytree.cgi?id=150226&hide_resolved=1</a>	<a href="#t14:17" class="time">14:17</a>
mclasen	ok	<a href="#t14:17" class="time">14:17</a>
wwoods	to be fair, there's a good 15 bugs that are on the ppc64 blocker	<a href="#t14:17" class="time">14:17</a>
wwoods	but yeah I'll do some triage and make some noise	<a href="#t14:18" class="time">14:18</a>
wwoods	and we'll re-evaluate for slip next week	<a href="#t14:18" class="time">14:18</a>
f13	I was thinking of the Test4 blocker bug that we created but didn't advertise.	<a href="#t14:18" class="time">14:18</a>
wwoods	(i.e. post-summit)	<a href="#t14:19" class="time">14:19</a>
wwoods	f13: ah, that one	<a href="#t14:19" class="time">14:19</a>
notting	wwoods: we'll reevaulated for slip after we supposedly freeze?	<a href="#t14:19" class="time">14:19</a>
wwoods	IIRC all of those that are still open should be on the FC7Blocker list	<a href="#t14:19" class="time">14:19</a>
wwoods	notting: I thought we weren't freezing yet?	<a href="#t14:19" class="time">14:19</a>
jwb	we're slushed	<a href="#t14:20" class="time">14:20</a>
notting	is we or ain't we? if we aren't freezing on thursday, we should probably announce it	<a href="#t14:21" class="time">14:21</a>
f13	we're slushy now, supposed to freeze on Thursday	<a href="#t14:21" class="time">14:21</a>
f13	I can't feel good about deep freezing only a couple days after a merged repo shows up.	<a href="#t14:21" class="time">14:21</a>
wwoods	I'm not too pumped about that either	<a href="#t14:22" class="time">14:22</a>
f13	and I hate slipping for an undeterminate amount of time.	<a href="#t14:22" class="time">14:22</a>
f13	My vote.  Yes slip, at least a week.  State that merger took longer/more effort than panned for, need more time for rawhide to settle down a bit and maintianers to get their builds in.	<a href="#t14:23" class="time">14:23</a>
jwb	+1	<a href="#t14:23" class="time">14:23</a>
f13	will re-evaluate blocker status next week and decide on further freezes for blocker reasons.	<a href="#t14:23" class="time">14:23</a>
jwb	at least a week	<a href="#t14:23" class="time">14:23</a>
wwoods	Yeah I'd really like at least a week after we get the first post-merge rawhide out	<a href="#t14:24" class="time">14:24</a>
wwoods	we're still pretty sure that will happen this week?	<a href="#t14:24" class="time">14:24</a>
jwb	and by "rawhide" you mean a real "rawhide" right?	<a href="#t14:24" class="time">14:24</a>
f13	mumble mumble	<a href="#t14:24" class="time">14:24</a>
notting	wwoods: one way or another	<a href="#t14:24" class="time">14:24</a>
* jeremy is okay-ish with a slip (well, as okay as can be expected :-)		<a href="#t14:24" class="time">14:24</a>
wwoods	mumble, handwave, mumble	<a href="#t14:24" class="time">14:24</a>
wwoods	but yeah, one way or another I would dearly love to have something testable by thursday so we can freeze next thursday (and so we can bang the drums about some blockers)	<a href="#t14:25" class="time">14:25</a>
wwoods	so yes, slip a week on the freeze and re-evaluate blockers next week	<a href="#t14:27" class="time">14:27</a>
* notting wanders off to go postal		<a href="#t14:28" class="time">14:28</a>
f13	ok.  We have our normal Monday meeting next week, perhaps we should have a second one on Wed?	<a href="#t14:28" class="time">14:28</a>
jwb	so we remain in slush?	<a href="#t14:28" class="time">14:28</a>
f13	jwb: yes.	<a href="#t14:28" class="time">14:28</a>
warren	+1 slip freeze, too many people gone this week	<a href="#t14:28" class="time">14:28</a>
jwb	k	<a href="#t14:28" class="time">14:28</a>
jwb	f13, do you want me to take a crack at the "before->after" doc?	<a href="#t14:29" class="time">14:29</a>
f13	jwb: sure!	<a href="#t14:29" class="time">14:29</a>
f13	brb	<a href="#t14:29" class="time">14:29</a>
jwb	f13, ok i'll try	<a href="#t14:30" class="time">14:30</a>
jeremy	jwb: would be much appreciated	<a href="#t14:30" class="time">14:30</a>
jwb	i'll email rel-eng with a link once i have something	<a href="#t14:30" class="time">14:30</a>
jeremy	and are we going to continue forward with cvs branching on thursday?	<a href="#t14:30" class="time">14:30</a>
jwb	jeremy, i'd like to get a "real" rawhide working before we branch...	<a href="#t14:30" class="time">14:30</a>
jeremy	jwb: I'd love to wait until after we have a release to branch... but the overwhelming consensus seems to be that it's causing more problems than it's helping :/	<a href="#t14:31" class="time">14:31</a>
wwoods	i get the feeling developers are getting really antsy about the freeze	<a href="#t14:32" class="time">14:32</a>
f13	yeah, freezes suck	<a href="#t14:32" class="time">14:32</a>
jwb	jeremy, but if we branch and we don't have the f8 rawhide working people will be more pissed	<a href="#t14:32" class="time">14:32</a>
f13	getting a release out is _hard_ and somepeople don't like to play along.	<a href="#t14:32" class="time">14:32</a>
f13	jwb: yeah, we're not going to have a post-fc7 rawhide until fc7 goes gold.	<a href="#t14:32" class="time">14:32</a>
notting	so	<a href="#t14:33" class="time">14:33</a>
notting	1) who wants to go through dist-fc7 but not f7-final stuff and ping people?	<a href="#t14:33" class="time">14:33</a>
notting	(with buildroots, this is important)	<a href="#t14:33" class="time">14:33</a>
notting	2) do we want to shunt stuff to not-dist-fc7?	<a href="#t14:33" class="time">14:33</a>
jwb	f13, so we're not going to open the gates for dist-fc8 until f7 releases?	<a href="#t14:33" class="time">14:33</a>
f13	should be simple enough to get packages and nvr compare them.	<a href="#t14:33" class="time">14:33</a>
f13	jwb: goes gold, not necessarily released.	<a href="#t14:33" class="time">14:33</a>
jwb	ok, then why can't we branch at that point?	<a href="#t14:33" class="time">14:33</a>
f13	there is a lag between gold -> open.	<a href="#t14:34" class="time">14:34</a>
f13	jwb: some people feel that's _way_ too late.	<a href="#t14:34" class="time">14:34</a>
jwb	wtf is the difference?	<a href="#t14:34" class="time">14:34</a>
f13	but then again, these are the same people that don't see the value in rawhide matching what we want to ship during those times too.	<a href="#t14:34" class="time">14:34</a>
jwb	there is no rawhide at the moment.  that's my whole point	<a href="#t14:34" class="time">14:34</a>
f13	jwb: we can branch and allow them to build, and they could get to the "hot" repos, but it wouldn't go out in rawhide.	<a href="#t14:34" class="time">14:34</a>
jeremy	jwb: people want a place to be able to commit post-f7 things	<a href="#t14:34" class="time">14:34</a>
jwb	jeremy, hrm...	<a href="#t14:35" class="time">14:35</a>
jwb	k	<a href="#t14:35" class="time">14:35</a>
nirik	one issue with the devel freeze is that maintainers of formerly extras packages are pushing fc5/fc6/devel, and so when their devel package isn't in f7, EVR is likely to be broke...	<a href="#t14:35" class="time">14:35</a>
f13	nirik: yep, something that they have to be aware of	<a href="#t14:36" class="time">14:36</a>
f13	merging doesn't come for free, there is give/take on both sides of the merge.	<a href="#t14:36" class="time">14:36</a>
nirik	indeed, but I think there are some already oblivious ones. I guess we can just do the best we can.	<a href="#t14:36" class="time">14:36</a>
f13	The desire to have more control over the packages that are part of the distribution means that they have to care about the whole distribution.	<a href="#t14:36" class="time">14:36</a>
* f13 tries to calm down his crankyness.		<a href="#t14:37" class="time">14:37</a>
nirik	agreed, but I think some of them just need to be educated about it.	<a href="#t14:37" class="time">14:37</a>
f13	Ok, we're at over 1.5 hours here, is there anything else that needs to be talked about now?	<a href="#t14:37" class="time">14:37</a>
* jwb has to go to a real-life meeting		<a href="#t14:37" class="time">14:37</a>
wwoods	I'm good for now. plenty to do for all, I think	<a href="#t14:38" class="time">14:38</a>
f13	ok, ***** end *****	<a href="#t14:38" class="time">14:38</a>

Generated by irclog2html.py 2.3 by Marius Gedminas - find it at mg.pov.lt!