ReleaseEngineering/Meetings/2008-dec-01

From FedoraProject

Jump to: navigation, search

Contents

Fedora Release Engineering Meeting :: Monday 2008-12-01

Fedora 11 Schedule

Other Plans for Fedora 11

  • build out documentation so we can grow Release Engineering beyond current members
  • signing server
  • more automation around check-ins and builds

Fedora 11 Naming

  • f13 to check with Josh Boyer (jwb) on running the process

IRC Transcript

f13 ping: notting jeremy rdieter_away spot lmacken poelcat wwoods warren 10:00
* notting is here 10:00
* lmacken 10:00
* warren here 10:00
jeremy hi 10:00
* poelcat here 10:03
f13 oh sorry 10:06
f13 got distracted 10:06
-!- f13 changed the topic of #fedora-meeting to: Fedora releng - F11 Schedule 10:06
f13 .rel 843 10:06
zodbot f13: #843 (Draft Fedora 11 Schedule) - Fedora Release Engineering - Trac - https://fedorahosted.org/rel-eng/ticket/843 10:06
* f13 wishes wwoods was here 10:06
notting as i said in e-mail to poelcat, the latest proposal (http://poelstra.fedorapeople.org/schedules/f-11/f-11-draft.html) seems reasonable 10:08
poelcat f13: i'll make other wording changes you suggested in ticket 10:09
f13 I added a comment in the ticket about the tail end of the schedule 10:09
* poelcat votes for "send it to FESCo!" 10:09
wwoods PATCHOW 10:09
f13 wwoods had been talking about radical changes he'd like to see at the tail end too, but unfortunately he's not here 10:09
* wwoods here now 10:09
f13 OR IS HE!?! 10:09
* wwoods appears in a poofy cloud of flour 10:09
f13 10:10
wwoods smokebombs smell funny. anyway: scheduling magic 10:10
wwoods There aren't any radical changes to the schedule per se - at least not against the version I'm looking at 10:11
f13 wwoods: I thought you were talking about doing away with Preview? 10:12
wwoods The only real change was to make the final freeze more final (and more frozen) by demanding final builds *before* then 10:12
wwoods emphasizing that by calling Preview "RC1" 10:13
wwoods and rejecting any features that wouldn't be fully complete by then 10:13
wwoods I don't think that's a *radical* change though 10:14
f13 so I'm with you on making the final freeze more strict 10:14
f13 and I think one way to do that is part of my suggestion, moving the RC compose date up to the 5th rather than the 12th 10:15
wwoods I'm undecided on renaming (since I doubt that Preview will actually be a viable RC) 10:15
f13 I'm not comfortable calling Preview an "RC" though 10:15
wwoods yeah 10:15
warren Perhaps we should create a RHEL-like exceptions process, and make Beta the point where we get more strict? 10:15
f13 unless we actually made it an RC 10:15
f13 beta is really really early to get like that 10:15
warren That might beat people to get stuff done long before the hard deadline of RC. 10:15
f13 and there is no way I'm going to support ack-flags and cvs blockage for Fedora 10:16
wwoods yeah we're not doing that. 10:16
f13 warren: currently we state that your features have to be /testable/ by Beta, not perfect 10:16
warren I wouldn't ever suggest cvs blockage 10:16
f13 we give you from beta to final freeze to fix your bugs found in testing 10:16
notting f13: the 5th? that's not in the trac ticket 10:17
f13 oops 10:17
f13 sorry, the 12th 10:17
wwoods we need a spec by Alpha and something that (mostly) meets spec by Beta. If it doesn't fully meet spec at the *freeze* for PR, it gets booted 10:17
f13 wwoods: I'd rather have the 'boot' date a week prior to final freeze 10:18
wwoods right - the FESCo meeting before each freeze decides what will actually make it into the release 10:18
f13 wwoods: that gives us a week to enact the contengency plan without having to do a bunch of tag requests 10:18
wwoods if something gets booted, it gets booted before we freeze, and then we freeze without it 10:18
f13 what days are fesco meetings? 10:18
poelcat f13: wed 10:19
notting wednesdays, ATM 10:19
wwoods and since freezes are on tuesdays, that means it's a week 10:19
f13 ok, so that's one day less than a week, but I can live with that 10:19
wwoods but yeah, just in case the meetings move or whatever, let's be clear - the fesco meeting the week *before* the freeze is the deadline 10:20
wwoods anything that's not meeting the requirements at that moment *will* be dropped 10:21
f13 s/dropped/will have the contengency plan enacted/ 10:22
wwoods (unless it's definitely 100% sure going to be ready before the freeze) 10:22
f13 of course this means we need to have valid contengency plans 10:22
wwoods yes - assume that "dropping" a feature means that we follow the contingency plan 10:22
f13 other than "wait longer" 10:22
wwoods sometimes the contingency plan is just "don't tag these packages" or "don't add to comps" or whatever 10:23
warren or "do nothing, just don't advertise it on the feature list" 10:23
wwoods right 10:23
wwoods "dropped" sounds a lot more harsh, I should try to avoid it 10:23
wwoods "deferred" might be a better choice 10:24
f13 wwoods: just to recap, there are no date changes you'd recommend for the schedule? 10:26
wwoods no, I think being stricter about what changes we allow is sufficient 10:26
wwoods a month in Final Freeze should be plenty, so long as we stick to the Final Freeziness of it 10:27
f13 sadly it's going to result in another 300+ 0-day update set 10:28
f13 ok, is there any other input on the schedule? 10:29
wwoods it would be nice if we had some subset of the distribution - some Core part of Fedora - that was subject to the stricter freeze requirements 10:29
f13 Proposal: accept poelcat's draft and push to FESCo for approval 10:29
poelcat +1 10:30
f13 wwoods: I've seen that asked before, but where would you draw the line, how would you enforce it, and why couldn't we just do that anyway with what we accept/reject? 10:30
notting +1 10:30
spot +1 10:30
f13 (+1 to the proposal btw) 10:30
f13 warren: wwoods: jeremy: lmacken: rdieter_away: votes? 10:32
lmacken +1 10:32
wwoods +1 10:32
f13 well, I think that's enough to pass, so poelcat consider it passed. 10:34
poelcat f13: i'll forward to bpepple for inclusion on FESCo's agenda this wed 10:34
f13 ok. 10:36
-!- f13 changed the topic of #fedora-meeting to: Fedora Releng - F11 plans 10:36
f13 This should just be a quick braindump of what we'd like to work on for F11. 10:36
f13 I'd like to see the signing server take shap, mitr is working on that 10:36
f13 I'd like to start working on a messaging server as a communications medium for the basis of doing reactive QA testing 10:37
f13 (IE react to a cvs checkin, a package build, an update request, a rawhide push, etc...) 10:37
f13 I'd like to see an effort to gather up community ran scripts, fix them up as necessary for efficiency, conformity, and put them in the releng git, as well as setup scheduled runs of the scripts in Fedora infrastructure 10:38
notting scripts for...? 10:38
f13 I've got some bugs to fix in pungi, as well as helping out with a re-write of the anaconda compose scripts 10:38
f13 notting: there is broken deps checking, source url checking, conflicts checking, etc... 10:39
f13 many of these can eventually be turned into reactive tests as mentioned above, but in the meantime I think it would be worth getting these ran in some "official" capacity 10:39
notting i'd like to see more push-button release making, but i don't think i really have much time to work on it 10:40
f13 oh yeah 10:40
f13 documentation 10:40
f13 I'd like to borrow some docs writers time to help me and others beef up our documentation 10:41
f13 both from a SOP "how we do things" pov as well as a more maintainer appropriate "this is what we do and why" 10:41
f13 notting: push-button release making is going to be pretty hard 10:42
f13 but I can spend some time writing more scripts to do what I've been doing by hand 10:43
f13 it's just hard to script things across multiple systems 10:43
f13 ok, nobody else has any goals for F11? 10:49
notting 'ship it' 10:49
poelcat f13: was there anything left over from the F10 "things to do" ticket? 10:50
f13 poelcat: pardon? 10:50
poelcat didn't we have a ticket of F9 post mortem items for F10? 10:51
f13 yeah, they were kind of hand-wavy. 10:52
f13 I'll review the ticket and see if anything is still relevant or possible. 10:52
jeremy f13: while not an easy goal, it'd be nice to get more "outside" participation in rel-eng 10:53
f13 jeremy: yeah, I was hoping that our drive to bring in scripts would bring on some of the authors of those scripts to the team 10:54
f13 the first question to answer though is what would we want these people doing 10:54
lfoppiano hi 10:55
jeremy if we got to nirvana? actual building of release trees could move between people for various stages of the release and releases instead of it always being on your plate 10:55
f13 jeremy: yeah, that would be nice. I'm just worries about the traning and such aspect leading up to that 10:57
f13 I'd like to get more people looking after overall tree health and fixing things up there 10:57
f13 as well as script writing and such 10:58
f13 that could grow into more responsibility, and in many cases can be done without handing over sysadmin-releng rights 10:58
* jeremy is unconvinced that tree health is really the "duty" of rel-eng... it ends up being done by people who happen to be in rel-eng just because we're suckers ;-) 10:59
f13 well, sure I'd love to do less work, but then the work just wouldn't get done 10:59
jeremy but we have this very very large small-numbers (or arguably single) point of failure in terms of getting releases out, updates pushed, etc 11:00
notting first step is the aforementioned docs 11:00
jeremy notting: docs help. docs don't motivate people. 11:01
notting signing server also helps 11:01
f13 yeah, I'm painfully aware of the single point of failure 11:01
notting jeremy: it's the first step though. without docs, even motivated people are unable to help in many cases 11:01
f13 docs + people + trust = salvation 11:02
jeremy notting: motivated people can and will muddle through without docs. it's maybe not as fast or as efficient, but it's sometimes more fun for the do-er ;-) 11:03
notting jeremy: muddle through... to push updates? 11:03
notting (we're drifting off-topic) 11:04
jeremy notting: even there, yes. it starts with rough email that says this is basically what I do, you can figure it out 11:04
f13 I think each step is equally difficult 11:04
f13 starting with finding people who are A) motivated, and B) trustworthy, since right now you'd get keys to the castle 11:04
jeremy f13: now that there's separate keys per release and updates-testing, the keys to the castle are starting to be segregated 11:05
f13 to some extent 11:05
notting jeremy: if those you trust do something silly with the key, it's still a huge PITA 11:06
jeremy notting: and if those with access to any of the sysadmin groups do something silly with their ssh keys, it's still a huge PITA. 11:06
* nirik would be happy to help push updates, but notes that there is a C) enough time in the day to do all the fun things out there to do. ;) 11:06
jeremy we have got to get beyond the "oh dear god, the world could end!!!" mentality or we're always going to push people off and come up with excuses that we can't get more people really involved 11:08
jeremy anyway, </soapbox-that's-been-bugging-me-a-little-recently-and-needed-to-get-off-my-chest>  :-) 11:08
f13 jeremy: I think it's a valid point, and something we should look at for F11 11:09
notting jeremy: exactly, which is why we're pushing people off by ... documenting what we do so people can follow it, getting a signing server ready so we can automate more things/spread signing around, and... huh? 11:10
jeremy notting: except that we never get the time to actually *do any of those things* because the day to day sucks up all available time 11:11
jeremy we've been talking about working on the signing server "for the next release" for exactly how many releases now? 11:11
notting so, you're saying 'why make plans, we'll fail to execute on them?' 11:12
notting i really don't see what your angle is here 11:12
f13 jeremy: except this time we outsourced the signing server work to somebody who's explicitly tasked with doing it 11:12
f13 jeremy: and now we're getting results 11:12
jeremy notting: I'm saying planning for 28 hour days doesn't do anyone any good. you have to either a) just go ahead and start getting external people involved with the day to day to help make time for the other things, even if it's not the perfect situation or b) outsourcing some of the "new" things is also valid 11:15
poelcat how about starting with something as simple as tag requests? I see the tickets come in and would love to help from time to time 11:16
poelcat but there are no docs or simple guidelines on what to do 11:16
jeremy notting: my angle is that this meeting and this group is the same cabal it's always been. we've seen massive growth around people helping mmcgrath out with sysadmin-y tasks. we badly need to do the same with rel-eng if we're going to scale at all and not just burn all the candles at both ends 11:17
poelcat granted maybe you'd argue that is why I shouldn't help because what to do should be obvious :) 11:17
jeremy poelcat: there actually is some documentation of that one :-) 11:17
jeremy http://fedoraproject.org/wiki/ReleaseEngineering/SOP/BuildRootOverrides 11:17
notting jeremy: ... and i still state that the first step is *actually documenting what gets done*, because if someone working 28-hour days doesn't have time to do XYZ, they also don't have time to walk someone else through it when they're muddling along 11:19
f13 that's a lot of why sysadmin grew so much, they had good docs to cover what they did 11:20
jeremy f13: the docs came as a result of trusting for the best and giving some people some responsibility 11:20
jeremy because then there was time to write the docs 11:20
f13 ok, we're obviously going in circles here. We need both, we should shoot for both. 11:21
f13 jeremy: I'd love it if you would take up the task of trying to recruit more people in 11:21
f13 but we do have to ahve a better story for what they can do once interested. 11:21
f13 ... and I think a good target for that would be shepherding scripts into the rel-eng git tree and finding a place in infrastructure to run them. 11:23
f13 and ticket triage. 11:23
f13 we're horribly over time though, is there anything else for today's meeting? 11:24
* jeremy has little to no thinking or breathing room for the next week or two. but will try to get something a little more fleshed out once he gets through this current stretch 11:24
f13 ok. 11:25
f13 thanks all! 11:25
* f13 has to jet for a bike ride before the weather sets in 11:25
notting oh wait 11:25
f13 notting: hrm? 11:25
* notting was going to ask about f11 names 11:25
f13 oh yeah, forgot about that. 11:25
f13 that's still jwb's task, even though he's "not in releng" 11:25
f13 at least, I think he still wanted to run that 11:26
f13 I'll ping him next time I see him to get that going. 11:26

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!