ReleaseEngineering/Meetings/2007-aug-06

From FedoraProject

Jump to: navigation, search

Fedora Release Engineering Meeting :: Monday 2007-08-06

Summary

  • Releasing on schedule Tuesday
  • May have limited mirror coverage
  • Torrents will be ready.

IRC Transcript

Idea.png f13 changed the topic of #fedora-meeting to: Fedora Release Engineering<a href="#t13:01" class="time">13:01</a>
notting*chirp chirp*<a href="#t13:02" class="time">13:02</a>
warrennotting, too loud.<a href="#t13:02" class="time">13:02</a>
f13jwb: spot: jeremy; notting: warren: poelcat: rdeiter: wwoods: roll call<a href="#t13:02" class="time">13:02</a>
* notting is here. all irc, no */callweaver?<a href="#t13:02" class="time">13:02</a>
* spot is here<a href="#t13:03" class="time">13:03</a>
f13all irc for now<a href="#t13:03" class="time">13:03</a>
* warren here<a href="#t13:03" class="time">13:03</a>
f13so F8 test1.<a href="#t13:05" class="time">13:05</a>
poelcathere<a href="#t13:05" class="time">13:05</a>
f13supposedly it's on the way to mirrors, or was very close to that earlier this morning<a href="#t13:06" class="time">13:06</a>
f13not very good feeling about releasing tomorrow without having any mirror support.<a href="#t13:06" class="time">13:06</a>
* jeremy is here<a href="#t13:06" class="time">13:06</a>
skvidalf13: no one uses test1 anyway :)<a href="#t13:06" class="time">13:06</a>
f13right, silly me.<a href="#t13:07" class="time">13:07</a>
nottingbeing synced by mirrors, or being synced so it can be synced by mirrors?<a href="#t13:07" class="time">13:07</a>
f13thoughts/opinions on giving mirrors a couple days to picked itup?<a href="#t13:07" class="time">13:07</a>
warrenwhat is the standard # of days?<a href="#t13:08" class="time">13:08</a>
f13notting: being synced so it can be synced by mirrors :(<a href="#t13:09" class="time">13:09</a>
warrenWill mirrors be upset if they are given only one day and the torrents open before many have synced?<a href="#t13:09" class="time">13:09</a>
f13warren: usually we give them at least 2 days.<a href="#t13:09" class="time">13:09</a>
warrenoh<a href="#t13:09" class="time">13:09</a>
f13<@arangel> jk_meeting: copying the sources-DVD<a href="#t13:09" class="time">13:09</a>
f13<@arangel> should be done soon<a href="#t13:09" class="time">13:09</a>
f13so I have no friggin clue how they do this.<a href="#t13:09" class="time">13:09</a>
f13apparently they rsync to a host that has an nfs mount, then they copy from that host /to/ the nfs mount<a href="#t13:10" class="time">13:10</a>
f13so there is multiple network transfers going on.  Why they can't just dump it directly to the netapp is a mystery to me, and my queries into it have mostly resulted in "you don't want to know why."<a href="#t13:10" class="time">13:10</a>
jwbi'm here<a href="#t13:11" class="time">13:11</a>
f13so to me, it looks like the only way we can do a release on a Tuesday is if we get IS to pick up the ticket before close of business on Friday, and maybe not even then.<a href="#t13:11" class="time">13:11</a>
nottingso, do we need to add extra delay in the schedule for handoff?<a href="#t13:12" class="time">13:12</a>
f13notting: well, shooting for thursday releases in general gives us that, but when we slip, we should just slip a week rather than 5 days.<a href="#t13:12" class="time">13:12</a>
jeremythat's why we switched to thursday to begin with, wasn' it?<a href="#t13:14" class="time">13:14</a>
nottingdo we really have a choice this time? i hate to push it back again, but if we have zero mirrors...<a href="#t13:14" class="time">13:14</a>
jwbhow bad is that?<a href="#t13:15" class="time">13:15</a>
f13jeremy: yeah, but I had small hope that we could have gotten the tree fixed and handed to IS in time for a tuesday splipped release.<a href="#t13:16" class="time">13:16</a>
f13would have too, if it wasn't for those pesky unsigned packages.<a href="#t13:16" class="time">13:16</a>
jwbi mean... it's test1.  do we expect massive numbers of downloads for test1 out of the gate?<a href="#t13:17" class="time">13:17</a>
jwbor would torrent suffice until the mirrors come online<a href="#t13:17" class="time">13:17</a>
skvidalesp since it is test1<a href="#t13:17" class="time">13:17</a>
jwbright<a href="#t13:17" class="time">13:17</a>
nottingwell, the mirrors tend to get annoyed if they're excluded<a href="#t13:17" class="time">13:17</a>
jeremythe big problem (historically) has been a) traffic makes it harder for mirrors to get at the release b) mirrors get grumpy at their local users going to the master to get it since they haven't gotten it yet<a href="#t13:17" class="time">13:17</a>
f13might be worth hitting up mirror-list-d for their opinions.<a href="#t13:17" class="time">13:17</a>
jwbi seem to recall those being the same reasons for not doing early torrent releases<a href="#t13:18" class="time">13:18</a>
* f13 too<a href="#t13:18" class="time">13:18</a>
jeremyjwb: the latter is, yes<a href="#t13:18" class="time">13:18</a>
jwbi would think that the mirrors would care a little less about test1<a href="#t13:19" class="time">13:19</a>
jwbemail mirror-list-d and see i guess<a href="#t13:19" class="time">13:19</a>
jwbwould be a shame to slip just for this though<a href="#t13:19" class="time">13:19</a>
jeremyyeah, I think let's just do it already.  as painful as it may be on the mirror side<a href="#t13:20" class="time">13:20</a>
f13guess I'll get to work uploading live isos to torrent (:<a href="#t13:21" class="time">13:21</a>
nottingbut certainly inform/notify/apologize to the mirrors<a href="#t13:21" class="time">13:21</a>
jeremynotting: +100<a href="#t13:21" class="time">13:21</a>
f13brb, fios install time.<a href="#t13:23" class="time">13:23</a>
wwoodsso who's gonna be writing release notes then?<a href="#t13:27" class="time">13:27</a>
warrenf13, sweet, I wish I could get FIOS where I will live.<a href="#t13:28" class="time">13:28</a>
wwoodsbut that means dealing with verizon.<a href="#t13:28" class="time">13:28</a>
warrencan't be that bad<a href="#t13:29" class="time">13:29</a>
jeremywwoods: thanks for volunteering? ;-)<a href="#t13:31" class="time">13:31</a>
wwoodshellfire<a href="#t13:32" class="time">13:32</a>
wwoodsnope, don't have time<a href="#t13:32" class="time">13:32</a>
blizzardf13: time to set up the arlington, ma wireless network?<a href="#t13:32" class="time">13:32</a>
f13blizzard: hee<a href="#t13:32" class="time">13:32</a>
f13blizzard: spot has fios too<a href="#t13:32" class="time">13:32</a>
blizzardf13: yeah, time to start sharing bandwith & costs<a href="#t13:33" class="time">13:33</a>
blizzardf13: *cough*<a href="#t13:33" class="time">13:33</a>
f13release notes?  You mean the release announcement.<a href="#t13:33" class="time">13:33</a>
f13blizzard: erm..<a href="#t13:33" class="time">13:33</a>
f13so I'm not going to be availbale for a few.<a href="#t13:33" class="time">13:33</a>
f13lets just do torrent and release and ask the mirrors for forgivness<a href="#t13:33" class="time">13:33</a>
spotblizzard: fios was cheaper than my cable bill in Chicago.<a href="#t13:33" class="time">13:33</a>
wwoodsannouncement. yes, that. the bit with the bullet points like "this is an alpha release. not all the features are here. It's wonky on ppc. it might not upgrade from F7 right."<a href="#t13:33" class="time">13:33</a>
f13(and maybe we can make it show up on a tier mirror quickly)<a href="#t13:33" class="time">13:33</a>
blizzardspot: I'm not sure I want verizon touching my special internet places<a href="#t13:34" class="time">13:34</a>
spotoh, and you trust comcast more?<a href="#t13:35" class="time">13:35</a>
blizzardrcn<a href="#t13:35" class="time">13:35</a>
warrenblizzard, how has RCN been?<a href="#t13:35" class="time">13:35</a>
blizzardit's fine<a href="#t13:35" class="time">13:35</a>
blizzardnot awesome<a href="#t13:35" class="time">13:35</a>
blizzardbut not bad, either<a href="#t13:35" class="time">13:35</a>
warrenblizzard, you have HD DVR?<a href="#t13:35" class="time">13:35</a>
blizzardno<a href="#t13:35" class="time">13:35</a>
* spot speed tested at 16 M/s<a href="#t13:35" class="time">13:35</a>
blizzardspot: I have the cheapest bandwith I can get<a href="#t13:35" class="time">13:35</a>
warrenlooks like FIOS isn't available in Somerville<a href="#t13:36" class="time">13:36</a>
wwoodsverizon charged my wife $1500 for night & weekend minutes when she had a free nights & weekends plan<a href="#t13:36" class="time">13:36</a>
f13I don't do verizon phone, nor copper<a href="#t13:36" class="time">13:36</a>
f13I'm willing to give fiber a try.<a href="#t13:36" class="time">13:36</a>
wwoodsafter two weeks of arguing with them and having to fax them their own godforsaken brochure to prove the point, they removed the phone charges but not the *late* fees<a href="#t13:36" class="time">13:36</a>
wwoodswhich then went to collections, bad mark on her credit report (which we have disputed), etc.<a href="#t13:37" class="time">13:37</a>
wwoodsI have similar stories for everyone I have ever known who has used verizon wireless<a href="#t13:37" class="time">13:37</a>
jeremywwoods: verizon wireless != verizon<a href="#t13:39" class="time">13:39</a>
jeremy(fwiw)<a href="#t13:39" class="time">13:39</a>
nottingwwoods: the wireless department hasn't progressed worse than general suckiness + dainbramage phone neutering for me<a href="#t13:39" class="time">13:39</a>
nottingthe phone people, on the other hand... RAGE.<a href="#t13:40" class="time">13:40</a>
warrenWith all the crappiness of Sprint, I've never had any problems like this.<a href="#t13:41" class="time">13:41</a>
blizzardor with t-mobile<a href="#t13:41" class="time">13:41</a>
blizzardwhich I use<a href="#t13:41" class="time">13:41</a>
blizzardand am pretty happy with<a href="#t13:41" class="time">13:41</a>
blizzardtheir customer service is awesome<a href="#t13:41" class="time">13:41</a>
* kanarip uses t-mobile too, happy with it<a href="#t13:43" class="time">13:43</a>
jeremydid we have any other agenda items for today?<a href="#t13:43" class="time">13:43</a>
nottingwho is *sending* the mail to the mirror list?<a href="#t13:44" class="time">13:44</a>
nottingand how do we fix the slip issue in the future?<a href="#t13:44" class="time">13:44</a>
jeremy(since I'm not even close to there in my mail)<a href="#t13:44" class="time">13:44</a>
nottingjeremy: btw: JINX! (re: ralink)<a href="#t13:44" class="time">13:44</a>
jeremywhat caused the slip?  was it just the dell thing?<a href="#t13:44" class="time">13:44</a>
nottingslip #1 (last week): dell, toshiba, etc. not booting (x86 setup rewrite-in-C fallout)<a href="#t13:45" class="time">13:45</a>
nottingslip/no slip today: didn't get tree handed off<a href="#t13:45" class="time">13:45</a>
f13which was caused by discovering non-signed packages.<a href="#t13:45" class="time">13:45</a>
f13sign-unsigned not perfect.  Warning.<a href="#t13:45" class="time">13:45</a>
nottingmore realistically... when was the last date we hit as scheduled? is it a different issue each time?<a href="#t13:45" class="time">13:45</a>
f13would have had a tree to hand off Friday morning, but alas had to respin all day friday.<a href="#t13:46" class="time">13:46</a>
jeremysigned packages has been an issue in the past<a href="#t13:46" class="time">13:46</a>
f13both kernel and signed packages have been issues<a href="#t13:46" class="time">13:46</a>
f13as has getting bits to the mirrors in time<a href="#t13:46" class="time">13:46</a>
f13the second slip is mostly my fault.<a href="#t13:46" class="time">13:46</a>
jeremythe technical things tend to change over time, although kernel comes up with regularity.  but no way to change that as long as we want to track the upstream kernel closely<a href="#t13:46" class="time">13:46</a>
f13I could have checked for signed packages at any point on Thursday but I completely forgot.<a href="#t13:46" class="time">13:46</a>
nottingiirc, there were some tree composition issues that would have possibly caused a slip if not for the kernel thing?<a href="#t13:47" class="time">13:47</a>
jeremyhow can we just generally make our signing more reliable and less prone to human error, though?<a href="#t13:48" class="time">13:48</a>
wwoodschecking signatures is really easy<a href="#t13:48" class="time">13:48</a>
wwoodsshould we be focused on checking the media that is produced, or ensuring that the input packages are signed?<a href="#t13:49" class="time">13:49</a>
nottingsimilarly, should we automate daily spins, etc?<a href="#t13:49" class="time">13:49</a>
jeremynotting: I lean towards yes.  as if we did and even looked at them occasionally, we'd be far better off for finding the things like "whoops, cd installs broke", etc<a href="#t13:50" class="time">13:50</a>
wwoodsit could help us work out some kinks in the workflow<a href="#t13:50" class="time">13:50</a>
poelcatf13: please let me know if there is are specific decisions and or summary(s) you'd like in the meeting recap<a href="#t13:51" class="time">13:51</a>
nottinghow do we do that? along the side of rawhide? something else?<a href="#t13:51" class="time">13:51</a>
* poelcat is semi-distracted coordinating shipment of fedora booth stuff to Linux World<a href="#t13:52" class="time">13:52</a>
jeremythe reasons we don't do isos for rawhide are a) space and b) compose time?<a href="#t13:52" class="time">13:52</a>
wwoodsright, both netapp space and mirror space<a href="#t13:53" class="time">13:53</a>
nottingjeremy: i'm wondering if they shouldn't just be a temporary .fp.o hosting<a href="#t13:53" class="time">13:53</a>
nottingi.e., not pushed to mirrors<a href="#t13:53" class="time">13:53</a>
warrennotting, what specifically?<a href="#t13:53" class="time">13:53</a>
jeremynotting: would be fine by me<a href="#t13:53" class="time">13:53</a>
nottingof course, that doesn't help *our* space issues<a href="#t13:54" class="time">13:54</a>
jeremyindeed ,but they just generally need resolving<a href="#t13:54" class="time">13:54</a>
nottingor are the rawhide trees 'good enough' for testing the tree sanity?<a href="#t13:55" class="time">13:55</a>
f13poelcat: Releasing on schedule Tuesday, may have limited mirror coverage.  Torrents will be ready.<a href="#t13:55" class="time">13:55</a>
poelcatf13: thanks<a href="#t13:56" class="time">13:56</a>
f13jeremy: see rfe for signing server, which can get a lot more of robust about making sure the signature and import and write actually happened.<a href="#t13:56" class="time">13:56</a>
f13jeremy: or in the mean time, more work on the /lovely/ sign_unsigned scripts<a href="#t13:56" class="time">13:56</a>
jeremynotting: if they're not, we should fix that :)<a href="#t13:56" class="time">13:56</a>
wwoodsdoing image composes won't really help with the signature problem, and boot.iso also triggered the kernel booting problem<a href="#t13:57" class="time">13:57</a>
wwoodsso.. it wouldn't have directly prevented either of those two things<a href="#t13:57" class="time">13:57</a>
nottingok, so, perhaps just get more people doing rawhide tests, and 'pay more attention to failures'? that seems cliched.<a href="#t13:57" class="time">13:57</a>
jeremywwoods: it would have fixed some of the other things we hit earlier in the freeze, though<a href="#t13:58" class="time">13:58</a>
jeremyand I think we just really need to pay a lot closer attention to "doesn't boot" bugs; even though they're incredibly painful to track down<a href="#t13:58" class="time">13:58</a>
nottingok, so the plan would be...?<a href="#t14:02" class="time">14:02</a>
nottingwwoods: is snake viable for daily rawhide sanity reports?<a href="#t14:02" class="time">14:02</a>
wwoodsnot yet - it lacks a scheduler and reporting mechanism (which are, technically, outside the scope of SNAKE, but anyway)<a href="#t14:03" class="time">14:03</a>
wwoodsI'm planning on working on SNAKE stuff next week<a href="#t14:03" class="time">14:03</a>
wwoodsI'd like to have it viable for daily reports by test2<a href="#t14:03" class="time">14:03</a>
wwoodsfailing that I'll hit KATE with a stout rod until it does the right thing<a href="#t14:04" class="time">14:04</a>
wwoodsand see about getting those reports public<a href="#t14:04" class="time">14:04</a>
wwoodsthis week is crazy administrative-type stuff that I've been putting off and setting up bugzilla (again)<a href="#t14:05" class="time">14:05</a>
f13I think regardless if we publish them anywhere, we should be doing weekly full composes<a href="#t14:10" class="time">14:10</a>
f13and that's really just bad on me for not doing them myself.<a href="#t14:10" class="time">14:10</a>
f13but we coul ddo them and toss them up on koji.fp.o/rel-eng/trees<a href="#t14:11" class="time">14:11</a>
f13just like we're doing weekly live isos on torrent<a href="#t14:11" class="time">14:11</a>
f13maybe we could do weekly install isos on torrent too<a href="#t14:11" class="time">14:11</a>
wwoodsI need to adapt SNAKE to deal with non-local trees (i.e. stuff that doesn't just land on the netapp)<a href="#t14:11" class="time">14:11</a>

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