18:25:23 #startmeeting Fedora Release Engineering 18:26:09 ping: notting wwoods lmacken jwb spot warren rdieter poelcat dgilmore 18:26:18 * poelcat here 18:26:26 no bill, he's on a plane I think. 18:26:33 no rex, he doesn't seem to be on IRC 18:26:48 no luke, he doesn't appear to be on IRC either 18:27:03 i'm here 18:27:04 yay 18:27:20 f13: http://poelstra.fedorapeople.org/schedules/f-12/f12-releng-updated-version-2009-07-13.txt 18:27:34 thanks 18:28:40 * wwoods hereish 18:29:57 alright 18:30:06 #topic F12 Schedule 18:30:53 A few of us just had a Fedora Talk and Gobby session to fine tune the F12 schedule. We've created a test compose prior to freezes, and tuned the blocker review days as well as the tail end of the schedule with go/nogo decisions and such 18:31:11 a /ton/ of work has gone into this schedule so this is last call for anything you find wrong 18:31:26 http://poelstra.fedorapeople.org/schedules/f-12/f12-releng-updated-version-2009-07-13.txt is the current iteration 18:31:29 there's no entry for "enable F12 updates" 18:31:43 that's a good point. 18:31:50 jwb: when would you like to start pushing F12 updates? 18:32:00 so i have a question first 18:32:14 does this account for "no frozen rawhide"? 18:32:31 no 18:32:44 that proposal has yet to be accepted, so this is the same business as usual 18:33:33 so rawhide is frozen from 9-22 until GA? 18:33:42 question: do you know how to see which video driver is running? 18:33:54 Ab3L, ask in #fedora 18:34:19 jwb: yeah, that should match up with previous releases, freeze just prior to "Preview" and remain frozen 18:34:51 f13, have we decided if/how we're going to do inheritance for updates before GA? 18:35:08 e.g. how are we going to avoid inadvertently breaking freeze for both the compose and buildroot 18:35:37 more koji tags 18:36:21 so we can safely enable updates at, say, Beta Release 18:36:45 yes 18:36:54 for some value of "safely" 18:36:59 hm. that might be too early 18:37:23 I think we're also missing the mass branch date here 18:37:30 yes 18:37:36 i was just looking for that 18:37:43 we had dropped af ew things to aid in concentration on the important dates we were discussion 18:37:47 this isn't a full schedule list 18:37:47 the two might want to go hand-in-hand 18:37:52 pre-branch is 2009-08-18 18:38:14 poelcat, that's allowable. we're talking about branching everything regardless 18:38:33 in the old style, we had pre-branching and then forced mass branching 18:38:36 jwb: okay, afraid we'd left soemthing out 18:38:54 I think we previously did forced mass branching just before PR 18:39:30 which is what now? beta? 18:39:38 beta 18:39:43 trying to find my announcement 18:40:56 so my concern with doing updates for f12 starting with beta is running into more XFCE situations 18:41:00 We did it at 4/15 for F11 18:41:09 jwb: yes, that is a concern. 18:41:23 * nirik looks up. wonders what he did. 18:41:28 nirik, nothing 18:41:39 nirik: we had packages in updates, that we also wanted in the final release 18:41:44 ah, right. 18:41:58 so we mass branched just before the "final freeze" 18:42:12 perhaps we disallow stable updates until 10-29 (or whenever we sync to mirrors) 18:42:19 and just allow updates-testing 18:42:33 so we'd be mass branching around 9/21 18:42:57 jwb: not sure I like that, leaves no way for 0-day updates for things 18:43:04 or pre-staged 0-day updates 18:43:11 sure it does... GA is 11-3 18:43:19 so we have 5 days 18:43:20 oh, sorry, my mistake 18:43:59 i was going with "RC is final" 18:46:42 but you're talking about 2 things though. A) when bodhi will accept requests, and B) when you'll push portions of those requests 18:47:47 yes. though A is mostly pointless without b... 18:47:56 yeah 18:48:25 well, if we think about it in terms of no-frozen-rawhide, at what point would we flip the switch to push things to 0-day updates as opposed to freeze breaks? 18:48:32 that'd be around the RC phase right? 18:48:46 think so 18:50:05 could just shoot for the same date now 18:50:38 f13, yeah. and in the interest of progress, we can just go with that now and revisit if needed later. it's somewhat orthogonal to the fundamental schedule 18:50:45 k 18:51:29 #agreed Mass branch will happen just prior (day before) to final freeze 18:52:23 #agreed bodhi will accept/push updates as of 2009-10-21 which is the RC phase for the final release. 18:52:43 #action poelcat will add such dates to the schedule 18:53:02 Was there anything else on the schedule? 18:54:04 guess not 18:54:17 #topic Orphan round up 18:54:31 We're supposed to do this near Feature freeze, which is coming up fast 18:54:37 warren did it in the past for us 18:54:48 is he not around? 18:55:21 I think I did it last time around 18:55:42 I can do it this time too 18:55:57 I'll need to send a list of the orphans a week prior to feature freeze 18:56:27 which would be 7/21, a week from tomorrow 18:56:51 #action jkeating will file a releng ticket to purge the orphans by feature freeze 18:58:11 looks like the script needs some work too, pkgdb changes likely 18:58:52 #topic FAD follow up 18:59:08 I really ahve to get no frozen rawhide pushed through, so I will be spending some time on that this week 18:59:30 I may have to call some more meetings to fine tune some of the missing data points 18:59:37 ok 19:00:40 I have nothing else for this 19:01:37 #action f13 will spend more time on FAD follow up this week 19:01:47 #action f13 will file a releng ticket to purge the orphans by feature freeze 19:01:52 #topic open floor 19:02:06 last update push took about a week 19:02:07 As you can see, we're using the new meet bot, this will help us with our meeting logging and reporting 19:02:12 for one reason or another 19:02:12 jwb: A WEEK?! 19:02:16 yeah 19:02:18 clarification: bodhi will accept/push updates as of 2009-10-21 == enable F12 updates ? 19:02:20 oh, just various issues 19:02:23 yeah 19:02:27 not a week to mash? 19:02:42 poelcat: yeah, that sounds right 19:02:45 correct, not a week to mash. longest repo mash time was 24 hours i think for f11-updates. i have one running rightnow 19:03:14 somewhat annoyingly, we can have a mash run for hours and then fail because koji pruned a signed RPM 19:03:21 which causes big delays 19:03:38 yeah, I thought we were going to back down the pruning a bit 19:03:43 leaving more than just one copy 19:03:49 er more than just the latest copy 19:03:57 guess that hasn't happened yet 19:03:59 #topic updates pushes 19:04:03 open floor question - where does critical-path controls for packages come into play? 19:04:06 ok, guess we need a ticket for that. 19:04:07 ah, sorry 19:04:23 #action f13 will file a ticket to remember to back down koji signed package pruning 19:04:35 f13, i'd almost rather have bodhi call write-signed-rpm on all RPMs before mash, or have mash do it once on error case before puking 19:05:06 jwb: if you'd like to work on that code, that's fine, however it would require that mash/bodhi/whatever run as a user with admin rights 19:05:09 which is a tad...s cary 19:05:12 plus cert management 19:05:16 bodhi already does? 19:05:20 considering it moves tags... 19:05:22 yeah 19:05:48 so really should be fairly simple. except bodhi doesn't do anything with signing at the moment 19:05:54 dunno if lmacken will like that or not 19:05:56 jwb: it shouldnt be pruning the latest ones 19:06:08 dgilmore, i don't believe it is 19:06:20 it's pruning older ones 19:06:21 dgilmore: it's not the latest 19:06:24 which is what is so frustrating 19:06:32 because I think it's a bug in bodhi that is making this harsh 19:06:50 jwb: then why is it being mashed? 19:06:50 bodhi is trying to re-push obsoleted updates into -testing I think 19:06:58 ok 19:07:14 or they're not getting properly obsoleted 19:07:15 f13, i hit it for more than just updates-testing 19:07:18 but I'd need more data 19:07:38 jwb: then something is going very wrong, and we shouldn't paper over it with kludges in bodhi/mash 19:07:46 do you have any examples we can look at? 19:07:50 sec 19:08:01 koji write-signed-rpm 4ebfc273 python-EnthoughtBase-3.0.2-1.fc10 19:08:12 that's the last one i ran. for f10-updates i believe 19:08:40 Thu Jul 2 08:07:25 2009: Tagged python-EnthoughtBase-3.0.2-1.fc10 with dist-f10-updates 19:08:40 Fri Jul 10 07:13:31 2009: Untagged python-EnthoughtBase-3.0.2-1.fc10 from dist-f10-updates 19:08:40 Fri Jul 10 07:13:42 2009: Tagged python-EnthoughtBase-3.0.2-1.fc10 with dist-f10-updates [still active] 19:09:05 jwb: that sounds like sign_unsigned did not write the signed rpm 19:09:32 odd that it was tagged on the second, then untagged/retagged today 19:09:39 er not today, the 10th 19:10:02 f13, and only 11 seconds apart... 19:10:29 perhaps that untag tells koji to prune it 19:10:29 jwb: that sounds like bodhi's doing 19:10:47 bodhi - 2009-07-03 19:50:31 19:10:47 This update has been pushed to stable 19:10:47 could be part of a bundled update? 19:10:50 afik there is nothing in koji that prunes sigs 19:10:56 bodhi seems to think it pushed to stable on the 3rd 19:11:00 dgilmore: yes there is 19:11:03 dgilmore, it's not the sigs. it's the signed RPMs 19:11:03 its all managed outside of koji code 19:11:09 dgilmore: the koji-gc scripts do it 19:11:18 well sure, pedantic, nothing "in koji" does it 19:11:26 f13: right but thats outsidenof koji 19:11:37 it's outside of the koji code base, but it's part of our koji deployment 19:12:02 f13: what i meant was nothing in koji-hub that goes and deletes signed rpms 19:12:17 Yeah, I don't think anybody claimed that there was 19:12:52 f13: 14:10 < jwb> perhaps that untag tells koji to prune it 19:13:06 ah 19:13:07 f13: i was just saying it doesnt 19:13:08 missed that 19:13:36 so we could use some insight from luke as to what happened with this update 19:13:57 jwb: any more that you find where we are missing sigs, we need to investigate them on a case by case basis. 19:14:08 could be bodhi bugs, could be sign_unsigned bugs 19:14:23 well, considering it's happening to at least one repo per push, that's going to take a while 19:14:32 fortunately, F9 updates are dead 19:16:38 right, but the point is we shoudln't be hitting these scenarios. We need to fix the causes, not cure the symptom 19:17:06 right 19:17:57 i had one epel package that did not have a written out signed rpm that broke a push so far 19:18:36 that's actually a normal error path 19:18:45 well, assuming it wasn't signed at all 19:19:53 it was signed 19:19:58 just not written to disk 19:20:06 ok, then sounds like what i'm seeing 19:20:20 probably falls under "signed_unsigned is a bloody piece of crap" 19:20:45 its either there or koji failing to write it 19:20:51 anyway, is there anything else regarding updates we need to talk about at this meeting? 19:21:01 i have nothing 19:21:12 was there any other action item we wanted jwb? 19:21:24 did I miss the critpath discussion (if any)? 19:21:32 not yet 19:21:39 f13, nope. just wanted to give a quick update 19:21:50 I think we're out of time this week, but can have a specific meeting for critical path if we want it. 19:22:18 * wwoods defers to skvidal here 19:22:19 but for the sake of... 19:22:30 #topic Critical Path 19:22:38 skvidal: you had mentioned something about it? 19:22:39 we can do it next week 19:22:43 ok 19:22:51 #agreed Critical Path will be talked about next week 19:22:57 #topic Open Floor 19:23:05 anything else? Will close meeting in 2 minutes 19:23:12 signing server? 19:23:21 mitr: on my to-do list this week 19:23:47 f13: I'll be happy to help as much as I can. 19:29:07 ok, thanks all! 19:29:17 #endmeeting