From Fedora Project Wiki

Fedora Release Engineering Meeting :: Monday 2009-02-02

Delay of Fedora 11 Alpha

  • multiple bugs found, and discovered as far back as the original freeze. Attempts to fix the bugs with newer trees brought in different bugs.
  • remaining issues with
    • iSCSI
    • iprutils
  • Go forward plan:
    1. rebuild ppc with newer iprutils
    2. delay Fedora 11 alpha until 2009-02-05 with approval from FESCo
    3. hold all other schedule dates

gcc 4.4 Mass Rebuilde

IRC Transcript

* notting is here 13:13
f13 ping: notting jwb lmacken wwoods rdieter poelcat spot warren 13:13
* jwb is here 13:13
* spot is here 13:13
* poelcat here 13:14
rdieter yo 13:14
* lmacken is here 13:14
* wwoods here 13:14
f13 neat. 13:15
::: f13 changed the topic of #fedora-meeting to: Fedora releng - ALPHA!!!! 13:15
f13 so... last week sucked. 13:15
f13 multiple bugs found, and discovered as far back as the original freeze. Attempts to fix the bugs with newer trees brought in different bugs. 13:16
jwb wait 13:16
f13 after a long weekend, I finally found a tree that seems to be "good enough" 13:16
f13 jwb: yes? 13:16
jwb did you figure out what the differences were between your compose environment and the official one that caused the bugs to be missed? 13:16
f13 no. 13:17
f13 I also didn't spend a whole lot of time on it, particularly because I had to blow away both the local and the remote one due to space issues 13:17
f13 anyway, what we have now is an alpha tag (f11-alpha-2) that has the contents from rawhide-20090130 + a newer nfs-utils package. 13:18
f13 so far, all the tests I've done have passed, including the critical nfs, getting past iscsiadm errors in gui mode, and getting past timezone selection in gui mode. 13:18
f13 The two outstanding bugs that I'm aware of that will likely hit a lot of people is UUID/LABEL mounting of /, and vnc installs 13:19
f13 vnc seems to have stopped working somewhere between the initial freeze and now, however I'm willing to trade vnc for working nfs and graphical installs 13:20
f13 particularly because we don't actually know what fixed a few of these issues, so we don't know what we coudl cherry pick back to an older freeze target to avoid vnc breakage 13:20
* warren here 13:20
jwb f13, any chance to include the .73 kernel? 13:21
f13 I don't think so, unless we want to slip longer. 13:22
f13 Given where we are today, if the rest of our critical path tests complete OK, we could actually ship on Thursday 13:22
f13 meaning only a 2 day slip. 13:22
* stickster reminds that Tuesday is usually the day for PR value 13:22
f13 that would mean releasing it to mirrors tomorrow, after I spend all day today staging bits around. 13:22
f13 stickster: Alpha is less of a PR target 13:22
f13 its not exactly suitable for press/reviews/masses, it's more of a reliable starting point for developers to get on the rawhide train 13:23
stickster In one way, yes... the target audience is really people who want to test the raw bits like developers. 13:23
f13 very few features are functional, no artwork, lots of broken things, etc, etc, etc,... 13:23
stickster However, we also have a mandate to get our features plenty of PR early on in the process. 13:23
f13 stickster: I'd happily trade 4 days of development time for some extra Alpha PR 13:24
stickster Just wanted to make sure rel-eng was aware of that issue 13:24
f13 particularly because we asked for a full 6 month schedule so that we could avoid slips :/ 13:25
stickster But if the week translates into slip time, then frankly I'd rather just get Alpha out there. 13:25
stickster I'd be much more concerned about the PR aspect were this the Beta. 13:25
f13 I would too 13:25
f13 but beta is a blocking freeze and we likely would have had more help finding issues 13:25
* stickster nods and sits down 13:26
wwoods I've got a fix for a ppc blocker that should be building shortly, but... I don't think we care about ppc that much 13:26
f13 wwoods: I still wasnt sure if we were producing ppc alpha images 13:28
f13 that's jwb's call 13:28
jwb f13, if you can mash the buttons, we can produce them. rawhide install last week at least worked for some subset of PPC 13:29
f13 also on the very scary list, live images. I've been trying to find a tree to make real installs work, that I haven't produced live images in a while from a rc tree to test. 13:29
* wwoods wondering about KVM status, reading scrollback 13:30
f13 wwoods: what's the nature of the blocker fix? (he asks has he watches the rest of the evening slip away) 13:30
f13 wwoods: all my kvm installs have gone fine 13:30
jwb f13, patching iprutils to not hang forever when iprdump starts 13:30
wwoods f13: new iprutils package 13:30
jwb i'm doing a local build here and i'll even test it 13:30
jwb unless you already tested it wwoods ? 13:31
wwoods yeah, I tested a build locally 13:31
wwoods WORKSFORME 13:31
wwoods but I'd like confirmation, esp. with something built from koji 13:31
jwb ok. then once this mockbuild finishes to make sure i didn't screw up something, i'll commit it and tag/build 13:31
jwb will let you know when that is done 13:31
f13 what happens if we don't have this? 13:32
jwb ppc hangs forever on start 13:32
jwb because iprdump isn't running as a daemon 13:32
wwoods and, for some reason, we install and run that on all ppc, regardless of whether they actually have the hardware 13:33
wwoods or want dumping\ 13:33
f13 did something recently change to cause it to start hanging? 13:34
jwb unknown. lack of boot.iso for a while didn't help there 13:35
f13 is there an easy workaround for this? 13:35
wwoods boot to runlevel 1, disable iprdump (or fix its initscript) 13:36
jwb single user, chkconfig iprdump off 13:36
f13 hrm. 13:36
jwb or do it via kickstart or someting 13:36
f13 iprutils is only ppc right? 13:36
wwoods yep. 13:36
f13 I could just re-compose ppc with that build in, and not have to touch anything else. 13:36
jwb hm 13:37
f13 jwb: now you want a kernel too 13:37
jwb you are psychic 13:37
jwb the oopses appeared to be non-fatal 13:38
wwoods I'm not sure if I'd done any ppc installs before now, or if F10 installed iprutils 13:38
jwb we could likely live without the kernel bump, but i dunno 13:38
f13 if it works to get to yum... 13:38
jwb it does 13:39
jwb well with the iprutils fix/workaround 13:39
wwoods I think probably I just hadn't done any ppc rawhide installs since iprdump got its new upstream (Nov. 21) 13:39
f13 ok, no kernel for you 13:39
jwb iprutils is building in koji right now 13:39
f13 ok. 13:41
f13 Proposal: re-spin ppc with new iprutils, no other changes. 13:41
warren I know nothing about this, so can't vote on it. 13:42
notting schedule impact is? 13:42
f13 notting: it'll take about 40 minutes to respin ppc 13:43
f13 no other arch needs to change 13:43
f13 so no significant schedule impact. 13:43
notting then +1 13:43
wwoods hrm. my ppc DVD gave me the iscsi asplode 13:44
wwoods f13, are those images still syncing or something? 13:44
wwoods anyway, +1, since the fix is fairly obvious and the worst possible outcome is.. ppc still doesn't work right 13:45
f13 wwoods: shouldn't be, what sha1 do you have? 13:45
wwoods much to the disappointment of all 9 users 13:45
wwoods f13: hang on while I sha1 the DVD 13:46
f13 7d769b9f52e4a70802bab5106f6aa7b480bd34c1 is what I have 13:46
wwoods it'd be quicker if you just gave me the timestamp from .discinfo 13:48
wwoods oh wait, it finished. yeah, that's the sha1 i have 13:48
f13 that would involve loopback mounting and... 13:48
f13 wwoods: thats.... unfortunate. 13:48
f13 ppc work around, use text. 13:49
jwb vnc install worked for me friday 13:49
wwoods f13: alias discinfo='isoinfo -x /.discinfo -i' 13:50
f13 ah 13:50
f13 iso-info ? 13:50
wwoods discinfo ~/[wwoods@metroid devel]$ rpm -qf $(which isoinfo) 13:51
wwoods genisoimage-1.1.8-1.fc10.i386 13:51
wwoods blerg. you get the point 13:51
wwoods f13: ppc GUI was assumed to be a no-op anyway 13:51
wwoods I'll try vnc/TUI 13:51
jwb yes, agreed 13:52
f13 -x isn't a valid option 13:52
jwb i don't even think i tried to boot into X on pc 13:52
jwb er, ppc 13:52
jwb and when the GUI install tried, it vomited 13:52
wwoods but I thought the iscsi "ValueError: need more than 0 values.." trace was gone 13:52
jwb presumably due to the SELinux thing 13:52
* f13 wonders why we have iso-info and isoinfo 13:52
f13 wwoods: no, those errors still happen, it's just not causing a traceback at least on i386/x86_64 gui 13:53
f13 wwoods: we don't fully understand /why/ we were getting a trace there to begin with 13:53
wwoods yeah, isoinfo (from genisoimage) not iso-info (from libcdio) 13:53
f13 stuuuupid 13:53
f13 alright so we have a plan for ppc 13:54
f13 I'm going to get on that, start making some jigdo images for the i386/x86_^4 trees and syncing them to the master mirror, so that I can clear up space to start making live images. 13:54
f13 I'll get some live images staged for testing at some point today 13:55
f13 Proposal: Finish testing current tree and live images, propose 2 day slip for alpha release on Thursday 13:58
poelcat f13: implied is that we hold to all other existing dates? 13:59
f13 poelcat: yes. 13:59
notting meh. what are the chances we'll have to re-revise on testing? 13:59
f13 notting: the known problems are fixed, it's the unknown problems, but we're getting pretty past the usual install methods, into the realm of "would be nice, but meh" 13:59
notting and, of course, the usual "how do we avoid this in the future?" 14:00
notting but that's a separate discussion. +1 to the proposal 14:00
warren just go ahead 14:01
* poelcat +1 14:01
f13 so I don't think there is a FESCo meeting before tuesday, question is how do we get an emergency vote? 14:02
poelcat f13: file a ticket? 14:02
f13 probably, and send out an email 14:02
bpepple poelcat: yeah, that would probably be the best bet. 14:02
jwb do we have a majority here? 14:03
jwb nirik, jds2001, dwmw2, sharkcz, ping 14:03
sharkcz jwb: pong 14:03
f13 ok, I"ll file the ticket. 14:03
* nirik is around. 14:03
::: f13 changed the topic of #fedora-meeting to: Fedora releng - open floor 14:03
f13 any last minute items? 14:04
* notting swaps hats, reiiterates his vote 14:04
jwb sharkcz, nevermind. i guess we're moving on :) 14:04
dwmw2 jwb? 14:04
jwb was going to see if we could call an emergency fesco vote, but *shrug* 14:04
f13 jwb: I'm pretty sure ramming this through a sudden IRC vote will piss off at least 2 people. 14:04
jwb f13, name the 2 14:04
poelcat f13: gcc4.4 mass rebuild ... when do we need to plan that on the schedule? 14:04
f13 poelcat: prior to beta. 14:04
jwb f13, and explain to me how they are involved in this in any manner 14:04
warren poelcat: still havent' decidded what the compiler flags and minimum arch will be 14:05
f13 poelcat: IIRC we're going to need a build for gcc4.4, rpm sha256, and font provides. 14:05
jwb warren, that can be changed later 14:05
notting jwb: not after the mass rebuild 14:05
f13 poelcat: they should all be done at the same time, rather than 3 different builds. 14:05
jwb notting, oh, i misread 14:05
jwb i thought it was jsut bringing gcc 4.4 in 14:05
poelcat Features/gcc4.4 14:06
warren jwb: part of the decision is to make the minimum arch i486 or i586 (as jakub suggested earlier) or maybe even i686 14:06
warren jwb: so that's a pretty big deal that needs to be decided 14:06
jwb yes yes, i'm aware of that. i said i misread 14:06
poelcat Features/ArchitectureSupport 14:06
f13 ok, if there is nothing else, I'll close the meeting. Lots of work to do. 14:12
f13 thanks all 14:12

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