From Fedora Project Wiki
Fedora Release Engineering Meeting :: Monday 2008-03-10
Beta Release
- Test plan: http://fedoraproject.org/wiki/QA/TestPlans/Fedora9Install is the test plan
- http://fedoraproject.org/wiki/QA/TestResults/TestSummaryTemplate is the "fill in the boxes" summary of results
- Discussion around what is working and not working--see IRC log for details
PreUpgrade
- rel-eng has agreed to maintain the file and it will be distributed on the mirrors
- pub/linux/fedora/releases/releases.list
Master Mirror Space
- after the board's talk with legal, we're clear to remove anything we want from the mirror.
- https://fedorahosted.org/fedora-infrastructure/ticket/421 is the ticket for the disk space
- Discussion about what releases need to be kept around and which ones can be deleted to free up space
- Potentially getting up to 2TB of space from HP
IRC Transcript
| -!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering Meeting - Roll Call | 13:02 | |
| * warren here | 13:02 | |
| * notting is here | 13:02 | |
| f13 | ping: notting warren jwb rdieter jeremy wwoods poelcat spot | 13:02 |
|---|---|---|
| * wwoods here | 13:02 | |
| rdieter | here | 13:03 |
| jwb | here | 13:03 |
| jwb | ish | 13:03 |
| * poelcat here | 13:03 | |
| f13 | I'm trying to chanage the wiki page to list the now adjusted UTC time | 13:04 |
| f13 | but I'm getting 500'd | 13:04 |
| f13 | ok, well | 13:04 |
| -!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - Beta Freeze | 13:04 | |
| wwoods | so.. rel-eng meeting adjusts for DST (i.e. it's tied to US Eastern time)? | 13:05 |
| f13 | Tonight is the beta freeze, for tomorrow's rawhide. | 13:05 |
| f13 | wwoods: yeah | 13:05 |
| f13 | dist-f9-beta exists, is continually updated with the latest dist-f9 stuff, and I've got signing writing out signed packages right now | 13:05 |
| f13 | rawhide tomorrow will likely have a bunch newly signed packages, but likely not all of them | 13:05 |
| f13 | I'll update dist-rawhide later tonight | 13:05 |
| warren | about what time will the cut-off be? | 13:06 |
| jwb | update it as in ? | 13:06 |
| f13 | dist-rawhide -> dist-f9 right now, tonight it'lls witch to -> dist-f9-beta | 13:06 |
| jwb | yeah, ok | 13:06 |
| f13 | dist-rawhide is what feeds rawhide composes. | 13:06 |
| jwb | yep, just making sure i understood | 13:06 |
| f13 | warren: 0700 UTC~ | 13:07 |
| f13 | warren: likely a bit before | 13:07 |
| * spot is here | 13:08 | |
| f13 | so lots and lots of test trees coming | 13:08 |
| f13 | RHEL5 selinux is thwarting my attempts at using rhel5 + mock to compose trees, so we might not be able to do them in PHX for beta | 13:09 |
| f13 | but I'll be keeping a test location in sync with the trees I compose | 13:09 |
| f13 | wwoods: you have us a testing grid right? | 13:10 |
| f13 | we should make an announcement regarding it and invite people to use rawhide to test things | 13:10 |
| wwoods | http://fedoraproject.org/wiki/QA/TestPlans/Fedora9Install is the test plan | 13:12 |
| wwoods | http://fedoraproject.org/wiki/QA/TestResults/TestSummaryTemplate is the "fill in the boxes" summary of results | 13:12 |
| jeremy | sorry I'm late... free lunch ran long :) | 13:13 |
| f13 | mmmmm | 13:13 |
| f13 | so.... | 13:15 |
| f13 | life is going to be busy for a bit. Anything else anybody would like to add? | 13:16 |
| f13 | We should keep in pretty constant contact with the state of the beta | 13:16 |
| jeremy | wwoods: any indicator of how we're looking, especially relative to last week? | 13:16 |
| jwb | ppc was mostly happy last friday | 13:16 |
| f13 | daily runthroughs of the blocker list, culling the non blockers, etc.. | 13:16 |
| f13 | jeremy: X is better, still some evdev oddities, but not bad | 13:16 |
| f13 | trees are composing pretty regularly | 13:16 |
| f13 | including ppc | 13:16 |
| f13 | perl landed today | 13:16 |
| jeremy | f13: yeah, that's my impression too. but I just wanted to make sure I wasn't missing anything | 13:16 |
| * jeremy saw perl too | 13:16 | |
| warren | I would really appreciate if somebody could review my ltsp package, been working constantly for more than 2 weeks getting that ready. | 13:17 |
| jwb | what does that have to do with Beta? | 13:17 |
| jeremy | also, if we find a place to put it (I'll talk to mmcgrath in a bit), I can put up a live image | 13:17 |
| f13 | I think the NetworkManager bits to reliably read system config files landed so we can flip NM on by default | 13:17 |
| warren | jwb, I was told it needs to go in before beta or it doens't go in at all. | 13:17 |
| jwb | why is that? | 13:17 |
| warren | feature freeze? | 13:17 |
| f13 | if it's a freature | 13:17 |
| jwb | it can go in. it just won't be called a feature | 13:17 |
| jeremy | f13: we should check with dcbw where that stands (NM) | 13:18 |
| wwoods | jeremy: stage2 stuff is still problematic. haven't tried an iso install today yet | 13:18 |
| f13 | jeremy: yeah, I got a text message on Friday/Sat saying that the bits landed upstream and that he was going to do a rawhide build | 13:18 |
| f13 | yes, stage2 worries me a lot | 13:19 |
| wwoods | stage2=XXX definitelly doesn't work yet | 13:19 |
| jeremy | wwoods: clumens should be around I think and is your best bet for fixage | 13:19 |
| wwoods | he's in targeted selection training | 13:19 |
| jwb | ? | 13:19 |
| jeremy | wwoods: "stay on target!" :) | 13:20 |
| * nirik wonders if the kvm vmware video finds no modes is known/filed/being looked at? | 13:20 | |
| jeremy | nirik: it's filed (and affects vmware vmware too afaik) | 13:20 |
| wwoods | gah. the internal mirror disappeared from the mirrorlist for rawhide for some reason | 13:20 |
| nirik | ok, sounds good. I should see if the 'root can't login on console' thing I just saw is filed too. | 13:20 |
| jwb | welcome to life for the rest of us wwoods | 13:20 |
| f13 | oh, also | 13:21 |
| f13 | syslinux doesn't seem to like recent kvm | 13:21 |
| jeremy | wwoods: I can try to take a look too if you have concrete reproducers of problems | 13:21 |
| wwoods | forget that! I don't wanna deal with slow mirrors like you suckas | 13:21 |
| wwoods | jeremy: sure | 13:21 |
| wwoods | oh. right-alt is still enter (or space) in anaconda. grumble grumble | 13:21 |
| wwoods | upgrade traceback... while looking for /etc/rpm/platform | 13:22 |
| wwoods | bluh. I'll take this stuff offline and report bugs and junk | 13:22 |
| f13 | ok. | 13:22 |
| f13 | we got works ahead of us | 13:22 |
| jeremy | but otherwise, would it really be a freeze? :) | 13:23 |
| f13 | bad day for selection training. | 13:23 |
| wwoods | basically: shit's rough but I'm hopeful | 13:23 |
| * jeremy at least feels like we're freezing on top of something more solid | 13:23 | |
| notting | mmm, frozen quicksand | 13:23 |
| jeremy | f13: and an entire bad week for me. woo! | 13:23 |
| f13 | nice | 13:23 |
| f13 | well, anything else on Beta? | 13:24 |
| notting | jeremy: punting redoing keymaps to match X? | 13:24 |
| jeremy | notting: yeah, I thought we decided that a few weeks ago | 13:24 |
| f13 | alright; | 13:26 |
| -!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - Open Floor | 13:26 | |
| f13 | Given the beta, I'm not focusing on anything else this week. Is there anything else anybody wants to bring up? | 13:26 |
| * warren falls through the open floor. | 13:26 | |
| * f13 watches warren fall through the hole in the roof | 13:27 | |
| wwoods | this might be an infrastructure thing instead, but | 13:27 |
| wwoods | the preupgrade feature requires a place to put a releases.list file | 13:27 |
| wwoods | which.. surprise! lists available releases | 13:27 |
| notting | a place... on the mirrors? on the wiki? in space? | 13:27 |
| wwoods | that's the question, really | 13:28 |
| wwoods | does something like that go on the mirrors, or do we just find some random place in the fp.o space to put it | 13:28 |
| wwoods | it's kind of the Official (machine-parseable) List O' Available Releases | 13:28 |
| f13 | we probably don't want it to be a SPOF | 13:29 |
| wwoods | and it'll need some (small amount of) rel-eng involvement to flip bits when releases are made public | 13:29 |
| f13 | so.. | 13:29 |
| f13 | pub/linux/fedora/releases/releases.list ? | 13:29 |
| wwoods | e.g.: add Fedora 9 when F9 is released | 13:29 |
| wwoods | https://fedorahosted.org/preupgrade/browser/dev/releases.list | 13:30 |
| f13 | ok | 13:31 |
| f13 | so yeah, I think dropping it in pub/fedora/linux/releases/ makes sense | 13:31 |
| wwoods | f13: works for me. but yeah - requires updating when releases are.. released. which is a releng-y task. | 13:31 |
| f13 | aight. | 13:33 |
| f13 | I have one other peice of business I forgota bout | 13:33 |
| -!- f13 changed the topic of #fedora-meeting to: Fedora Release Engineering - Master Mirror Space | 13:33 | |
| wwoods | f13: so - can I say that rel-eng has agreed to maintain the file and it'll be distributed on the mirrors? :D | 13:34 |
| f13 | wwoods: sure | 13:34 |
| wwoods | rad. thanks. | 13:34 |
| f13 | https://fedorahosted.org/fedora-infrastructure/ticket/421 is the ticket for the disk space | 13:34 |
| f13 | after the board's talk with legal, we're clear to remove anything we want from the mirror. | 13:34 |
| f13 | Which could be everything up to the Fedora 7 bits | 13:34 |
| nirik | Beta is gonna have jigdo bits, correct? | 13:35 |
| f13 | I'm still syncing stuff to spinning disks in Bostford but once that's done we can do some rm -rfing | 13:35 |
| f13 | nirik: yes | 13:35 |
| notting | f13: so, fc6 disappears completely? no archive? | 13:36 |
| f13 | I'm going to announce likely later this week the pending removal of the bits, temporary for now, ideally put back in archive.fedoraproject.org later with redirects from the old locations | 13:36 |
| f13 | notting: we'll want to put the bits back online eventually, we just don't have the resources to do so immediately | 13:36 |
| notting | do the mirror heads in rdu/tpa have similar space issues? | 13:37 |
| f13 | but we have to clear something off to handle 9's release with split media once again, and secondary arch content. | 13:37 |
| f13 | notting: I do believe so, they're all supposedly matched sets | 13:37 |
| notting | well, do we have space for 9 release w/o seconday arches? | 13:38 |
| notting | and is there a concrete plan for the future other than begging? | 13:38 |
| jeremy | notting: begging... with feeling ? | 13:39 |
| jwb | f13, which secondary arch content? | 13:39 |
| f13 | jwb: ia64 is the immediate need | 13:40 |
| f13 | they're ready to start having us pick up their rawhide droppings | 13:40 |
| jwb | cool | 13:40 |
| jwb | wasn't aware of that | 13:40 |
| jwb | are they all hooked into koji? | 13:40 |
| dgilmore | jwb: partly | 13:40 |
| jwb | explain? | 13:40 |
| dgilmore | jwb: I need to finish making things work | 13:41 |
| f13 | notting: Without scrubbing /some/ content we don't have room for 9 at all | 13:41 |
| dgilmore | jwb: they are queueing builds but not trying to reproduce the buildroot | 13:41 |
| f13 | notting: my future plan is that we only keep the active or active -1 releases on the netapps, and we archive things to an archive server with less costly space issues | 13:41 |
| jwb | dgilmore, and that's acceptable? | 13:41 |
| notting | f13: sure, but we don't have one of those yet :) | 13:41 |
| dgilmore | jwb: for now | 13:41 |
| jwb | wait, why? | 13:41 |
| f13 | notting: correct. We have some of the stuff for it, but it's being used for backups right now | 13:41 |
| f13 | jwb: because it's better than nothing | 13:42 |
| notting | f13: yeah, just hosting non-critical things without a concrete plan for space acquisition makes me nervous | 13:42 |
| f13 | I still think that reproducing the buildroot exactly is impossible | 13:42 |
| jwb | i've offered to send hard drives before | 13:42 |
| jwb | jeremy says that's pointless | 13:42 |
| f13 | notting: we have plans, and budgets, but they're in later quarters | 13:43 |
| notting | f13: did we not think we'd run out of space this fast? | 13:43 |
| notting | jwb: drives is sort of misleading. would need drives, and a box, in the colo, and some sort of redundancy... | 13:43 |
| jwb | yeah, hence pointless | 13:44 |
| f13 | notting: I don't think mmcgrath was aware that we were going to do split media for 9 again, and wasn't expecting to host secondary arches | 13:44 |
| notting | then... say no? | 13:44 |
| warren | we could do split media jigdo only... | 13:44 |
| f13 | notting: board already said yes. | 13:44 |
| warren | or all media jigdo only... | 13:44 |
| f13 | warren: I'd rather not | 13:44 |
| notting | f13: then tell the board they have their head up their ass and cough up some hardware | 13:45 |
| jwb | f13, what the board says and what we can physically accomplish are entirely different things | 13:45 |
| f13 | we /can/ accomplish it | 13:45 |
| f13 | there is no reason to keep Fedora Core 2 crap on the master mirror share | 13:45 |
| f13 | it's just silly | 13:45 |
| jwb | 1-6 can go | 13:45 |
| f13 | we can temporarily move it offline until the archives space frees up | 13:45 |
| jwb | why are we hosting it at all anymore again? | 13:46 |
| jwb | the old stuff i mean | 13:46 |
| notting | i'm just not sure i see the point of 'oh, we'll put it somewhere else later' | 13:46 |
| f13 | cleaning 1-6 off (basically the core/ directory) gives us ample space. | 13:46 |
| notting | after all, you're going to blow up their fc6 installs for 6 months anyway | 13:46 |
| notting | why would the users then care that you're putting it back | 13:46 |
| warren | blow up? | 13:46 |
| notting | yum doesn't like no baseurls existing | 13:46 |
| f13 | maybe they'll get the point that their release isn't supported anymore | 13:47 |
| warren | does removing 1-5 give us enough space? | 13:47 |
| notting | admittedly, they haven't seen updates for however long, but now they'll get loud errors | 13:47 |
| f13 | warren: depends on how fast other arches spin up | 13:47 |
| notting | i just don't like the ridiculous measures to support (IMO) a silly decision | 13:48 |
| warren | notting, +1 | 13:48 |
| jwb | arent' you on the board? | 13:48 |
| notting | jwb: doesn't mean people listen to me | 13:48 |
| f13 | notting: then please go talk to the board and spot, 'cause that's where the pressure is coming from | 13:48 |
| jwb | notting, i'm just wondering if you said the same during the board meeting | 13:48 |
| f13 | HP has already sent us a box to hold the bits on disks until we can get archives up and runing. | 13:48 |
| warren | <notting> and is there a concrete plan for the future other than begging? <-- This doesn't bode well. | 13:49 |
| f13 | concrete plans involve budgetary planning, which we're doing. | 13:49 |
| f13 | although a lot of the problem is finger pointing when it comes to "who is going ot pay for this" | 13:49 |
| f13 | notting: but please don't use releng as a method to further disagree with the board. I don't really like being put in that position. | 13:50 |
| jwb | why is that? | 13:50 |
| * warren wonders if we should split the ISO and package mirrors into different trees. | 13:51 | |
| f13 | jwb: we elect a board for a reason, they've made a decision, and we're doing the best we can to accomplish their decision. | 13:51 |
| f13 | warren: wtf are you talking about? | 13:51 |
| warren | It makes less and less sense to have multiple copies of the same thing. | 13:52 |
| notting | f13: it's the responsibility of rel-eng to give feedback to the board if what they ask for is nuts | 13:52 |
| f13 | it also makes less and less sense to carry really old content on the expensive master mirror. | 13:52 |
| jwb | i think we should ax the older releases regardless of the current discussion | 13:52 |
| jwb | at very least 1-4 | 13:52 |
| f13 | notting: well, I don't necessarily agree that it's nuts. I think the timing is a little off, but it's nothign we can't deal with. | 13:52 |
| f13 | jwb: 1-4 doesn't give us enough space IIRC | 13:52 |
| f13 | 1-5 gets us closer | 13:53 |
| f13 | 1-5 + the Fedora 7 Test [1234] may just about cover it | 13:53 |
| f13 | we may get away with keeping 6 on the master netapp | 13:53 |
| jwb | does that include extras 3-5? | 13:53 |
| notting | f13: this deletion is planned for... post-beta? | 13:53 |
| f13 | jwb: yes, that includes it. | 13:53 |
| f13 | notting: planned for as soon as I get the bits mirrored to the HP box in bostford, and send an announcement to the mirrors / users | 13:54 |
| f13 | could be before beta, but will probably be after. | 13:54 |
| f13 | notting: but if you can convince the board to choose a different course of action before then... | 13:56 |
| notting | f13: do you or don't you want to be cc'd? | 13:56 |
| f13 | as it stands, getting new mirrors bootstrapped is pretty tough given how much old content there is. | 13:56 |
| jwb | notting, please cc rel-eng | 13:56 |
| f13 | notting: sure, keep me cc'd, I'd like to be informed. | 13:56 |
| f13 | jwb: ++ | 13:56 |
| notting | i'd just feel *much* better if archives was actually live now. | 13:57 |
| notting | was there something that prevented the HP donation going somewhere it could be used in a public manner? | 13:57 |
| f13 | notting: rack space in PHX | 13:57 |
| f13 | == 0 | 13:57 |
| notting | and we can't send it to serverbeach? | 13:57 |
| f13 | no ability to put things outside the firewall in Bostford | 13:57 |
| f13 | serverbeach provides us machines, not the other way around | 13:58 |
| warren | notting, serverbeach does only dedicated, not colo. | 13:58 |
| f13 | we'd have to buy another machien from them, or more storage | 13:58 |
| notting | so, barring adding a cname for archives.redhat.com to elvis.redhat.com... | 13:58 |
| f13 | and getting IS to agree to putting another box in the colo | 13:58 |
| warren | how much space does elvis have? | 13:58 |
| notting | about 22G ;) | 13:58 |
| f13 | does elvis have the 400~ gigs we need? | 13:58 |
| f13 | s/colo/dmz/ | 13:58 |
| warren | notting, where is elvis anyway? | 13:58 |
| notting | somewhere in the rdu 4th floor lab, iirc | 13:59 |
| warren | f13, what do you mean by redirect? http redirect? | 13:59 |
| f13 | warren: yes | 13:59 |
| f13 | warren: if we can. | 13:59 |
| warren | f13, and do we allow rsync access to archive.fp for people who want to mirror it for no good reason? | 13:59 |
| f13 | warren: saves us from re-filing export approval for all the old releases. | 13:59 |
| notting | well, actually elvis is a nursing home in texas with jfk battling mummies. but that's besides the point | 13:59 |
| f13 | warren: it should. | 13:59 |
| f13 | well, nothing will be deleted before our next meeting | 14:01 |
| notting | f13: how much space will the in-6-months archives.fp.o be? | 14:01 |
| f13 | notting: that's a mmcgrath question | 14:02 |
| notting | and i suppose there's not really space on spins either | 14:02 |
| spot | notting: HP is working on donating much more storage to us | 14:04 |
| f13 | probably not that much space, but again, that's a mmcgrath question. | 14:04 |
| spot | i'm working with mmcgrath to find a home for it. | 14:04 |
| jwb | spins has about 84gig free | 14:04 |
| spot | about 2 TB | 14:05 |
| jwb | though i think mmcgrath recently had an upgrade for spins somewhere | 14:05 |
| notting | f13: dumb question - we want to have a written offer of some sort for F9 to handle some of these GPL issues. is there a way to beta test that without tying us into keeping the beta for however many years? | 14:05 |
| f13 | ... beta test a written offer? | 14:05 |
| jwb | that seems... bad. and also a question for fedora-legal | 14:06 |
| f13 | what do you want to test? | 14:06 |
| notting | f13: i.e., does including the first wording of this offer in the beta tie us to keeping the beta online for 3 years | 14:06 |
| * jwb <- not lawyer | 14:06 | |
| f13 | notting: word it in such a way that it doesn't. | 14:06 |
| notting | heh | 14:06 |
| f13 | I don't think you can just put "NOT!" at the bottom. | 14:06 |
| notting | mdomsch: are you in charge of that? | 14:06 |
| f13 | you can put a chucnk of text saying "This is where the written offer will be" | 14:07 |
| f13 | but I'm curious what you're trying to ... test | 14:07 |
| notting | the last i heard is that there was going to be something in the toplevel readme about the source offer | 14:07 |
| notting | and i don't want it to cause problems later | 14:07 |
| f13 | yeah, I'm not capable of answering that | 14:08 |
| mdomsch | I'm still working on it.. | 14:09 |
| mdomsch | won't be ready this week though | 14:09 |
| notting | f13: but whatever we do, you *do* want to make sure it doesn't apply to betas? | 14:09 |
| mdomsch | I need to get FSF legal (brett smith?) to ack that for GPLv2 we don't _have_ to provide physical media with source, if it's online | 14:09 |
| jwb | and alphas and rcs | 14:09 |
| mdomsch | GPLv3 fixed that particular madness | 14:09 |
| mdomsch | notting, right - we really only want that to apply to releases | 14:10 |
| mdomsch | where we can expect to press media | 14:10 |
| mdomsch | at least for now | 14:10 |
| jwb | mdomsch, why the v2/v3 distinction? | 14:10 |
| notting | mdomsch: and is this using gpl as the original source license, or gpl as the collective license. | 14:10 |
| mdomsch | there's a desire to allow GPLv2 3b) for everything - but we're far from there yet | 14:10 |
| mdomsch | notting, both... | 14:10 |
| mdomsch | jwb, v2 (in the FSF's opinion of the past) meant we had to put source on physical media for people if they wanted it, because we put the binaries on physical media | 14:11 |
| f13 | notting: that is correct. Making pre-releases stick around for 3 years will very very very quickly run us out of disk space | 14:11 |
| mdomsch | v3 says basically "putting them online is sufficient" | 14:11 |
| jwb | is there something preventing us from using gplv3 which has this issue fixed ? | 14:11 |
| mdomsch | jwb, all the v2-licensed code in the distro? :-) | 14:11 |
| jwb | oh, yes | 14:12 |
| notting | f13: sorry to drag things off-topic. anything else? | 14:12 |
| f13 | I got nothing, but a hungry belly | 14:13 |
| f13 | anybody else? | 14:13 |
| jwb | let's talk about spins | 14:13 |
| jwb | just kidding :) | 14:13 |
| f13 | thanks all | 14:15 |
Generated by irclog2html.py 2.3 by Marius Gedminas - find it at mg.pov.lt!