[07:32] <mojonixon> we'll I can duplicate it....  Looks like the install is complete but aptitude is zombied.
[07:32] <mojonixon> If I kill it's parent process, the install completes w/o problems....
[08:26] <tepsipakki> mojonixon: I've seen that too
[08:27] <tepsipakki> but only on certain machines (which are identical to others)
[08:27] <tepsipakki> I haven't seen it with edgy/feisty
[09:12] <mojonixon> I can duplicate the problem on a vmware machine on physical hardware under dapper.
[09:23] <tepsipakki> mojonixon: what package makes it hang?
[03:29] <cjwatson> evand: ok, there's a daily alternate build of gutsy up now
[03:29] <cjwatson> evand: kernel installation will break though - see the linux-meta upload I just did, which should unbreak that
[03:30] <evand> ok, thanks
[03:30] <cjwatson> oh, meh, udeb seeds are busted too
[03:30] <cjwatson> excuse me while I fix up a load of stuff
[03:30] <cjwatson> :-)
[03:30] <cjwatson> evand: do you have a seed checkout? (http://wiki.ubuntu.com/SeedManagement)
[03:31] <evand> haha
[03:31] <cjwatson> we need to update the Kernel-Version bits in the installer seed each time d-i is uploaded for a new kernel version
[03:31] <cjwatson> kernel ABI, rather
[03:31] <evand> negative, I'll take care of that now
[03:31] <cjwatson> you'll be able to write to it once you're in ubuntu-core-dev
[03:32] <evand> ok
[03:39] <cjwatson> ok, seeds are happy, linux-meta will take two publisher cycles
[03:47] <evand> cjwatson: link to the linux-meta upload? I'm having a hard time finding it.
[03:49] <cjwatson> evand: only just uploaded, so may not be visible yet
[03:50] <evand> ah, gotcha
[03:50] <cjwatson> evand: https://lists.ubuntu.com/archives/gutsy-changes/2007-May/002430.html
[03:51] <evand> thanks
[03:59] <evand> is the d-i daily build process documented anywhere?
[04:06] <evand> cjwatson: also, what was wrong with the udeb seeds?  Sorry if I'm being overly inquisitive
[04:08] <cjwatson> evand: build process> the place to start there is probably by checking out http://people.ubuntu.com/~cjwatson/bzr/cdimage/mainline/ and also each of the pieces in configs/devel once you've checked that out
[04:08] <cjwatson> there's a README in there with some information, and then RTFS :)
[04:08] <cjwatson> inquisitive> you cannot be overly inquisitive ;-)
[04:08] <cjwatson> udeb seeds> the Kernel-Version was set to 2.6.20-15, so feisty's kernel udebs were still being pulled in
[04:08] <evand> wonderful
[04:08] <cjwatson> onto the CD
[04:09] <evand> oh ok, that's where I got confused
[04:10] <mojonixon> aptitude hangs (zombie) after installing all packages...
[04:13] <mojonixon> sorry  that las message was a reply to one from about 7hrs ago...
[04:14] <tepsipakki> mojonixon: but the log should tell you which one it was installing (alt+F4)
[04:15] <mojonixon> actually the last package installed was openssh-server, but it is fully installed.
[04:17] <tepsipakki> ok, here it always hung when installing some package, and it was always the same package for that machine
[04:17] <cjwatson> yeah, it was right at the end of the process
[04:17] <cjwatson> somebody needs to strace it to figure out what's failing to reap it
[04:17] <cjwatson> probably strace -f -s 1024 the whole pkgsel process, dumping to a file on disk as it'll be a lot of output
[04:17] <tepsipakki> at least the counter fell behind every time
[04:17] <mojonixon> If I change the package list it still failes.
[04:21] <mojonixon> If you'll give me a clue as how to strace the pkgsel process, I'll do it and send the results.
[04:22] <mojonixon> I can get the machine back to that stage of the install fairly quickly...
[04:22] <cjwatson> one sec
[04:28] <cjwatson> mojonixon: so, you're going to have to be fairly quick, although not superman-speed
[04:28] <mojonixon> np....  I've got the command lines of the process tree leading up to the failure if that helps.
[04:29] <cjwatson> mojonixon: run through the install until the "Installing the base system" step starts
[04:29] <cjwatson> mojonixon: as soon as that starts, switch to tty2 and look through 'ps x' for the main-menu process, and get its process id
[04:29] <cjwatson> then 'strace -f -s 1024 -o /target/tmp/main-menu.trace -p PID' where PID is that process id
[04:30] <cjwatson> the next two things main-menu does after that point are apt-setup and pkgsel, so if you leave that running until the failure you should have an adequate trace
[04:30] <cjwatson> it is of course possible that it will go away under strace :-/
[04:31] <mojonixon> ok...  I'll add debconf/priority select low to the preseed and try  it...
[04:31] <mojonixon> Of course it will go away under trace :)
[04:33] <cjwatson> no, please don't use expert mode
[04:33] <cjwatson> I'm not sure what other subtle things that might affect
[04:34] <mojonixon> the failure happens the same either way.....
[04:34] <cjwatson> ok
[04:34] <cjwatson> in that case that would give you a better window, yes
[04:34] <cjwatson> you could just start the strace right before selecting "Select and install software"
[04:36] <mojonixon> ok...
[04:43] <mojonixon> ok, I gotta zombie...
[04:43] <mojonixon> I also got a [tcb table full]  message from strace.
[04:45] <mojonixon> ok, I  got 2.9MB of trace, where do you want it?
[04:48] <mojonixon> you should be able to get the trace here.... http://www.vircio.com/main-menu.trace.bz2
[04:50] <cjwatson> thanks, grabbing
[04:50] <mojonixon> please let me know what you're looking for so that I can follow along....
[04:52] <cjwatson> that's odd, I don't think that's complete
[04:52] <cjwatson> there's no mention of aptitude in there
[04:53] <cjwatson> well, no non-trivial mention
[04:54] <mojonixon> ok, I can try it again....
[04:56] <mojonixon> this is the command I used to generate the trace:
[04:56] <mojonixon> where 2051 is the pid of /usr/bin/main-menu
[05:03] <mojonixon> Actually aptitude is mentioned under process 2544
[05:04] <cjwatson> yes, but not relevantly
[05:04] <cjwatson> I'm looking for an execve of aptitudde
[05:04] <cjwatson> aptitude
[05:04] <mojonixon> ko...
[05:04] <cjwatson> as in, the process actually running
[05:05] <cjwatson> the tcb thing is very weird, and possibly an strace bug, which is less than convenient
[05:05] <cjwatson> try -f -F ?
[05:05] <cjwatson> (without the ?)
[05:06] <mojonixon> ok...  will do give me a sec.
[05:14] <mojonixon> of course it didn't fail this time....  I'll do it again....
[05:19] <mojonixon> I got a 329MB trace file though :)
[05:22] <cjwatson> sounds better
[05:22] <cjwatson> but yeah, would need it to fail
[05:23] <mojonixon> of course
[05:35] <mojonixon> interesting....  did a snapshot of the machine at the "Select and install software point"
[05:36] <mojonixon> Then, I continued on without strace and the install hung.
[05:36] <mojonixon> Now, I'm running strace -F f and the install works....  dohh...
[05:39] <mojonixon> Then, I reverted the machine to the snapshot and tried again w/o strace and it hangs.
[05:41] <cjwatson> annoying
[05:43] <mojonixon> If I use strace -F (without -f) it hangs but of course the trace is useless.
[05:55] <mojonixon> ok any suggestions, I cannot get it to fail with strace -F -f
[06:25] <mojonixon> how do I install the strace udeb?  I've got the following in my preseed.cfg but it doesn't work
[06:26] <mojonixon> d-i anna/choose_modules multiselect openssh-client-udeb, strace-udeb
[06:31] <blackskad> cjwatson: can I safely reject bug 68298 ?
[06:31] <blackskad> cjwatson: the initial reporter can't reproduce it anymore
[06:31] <blackskad> cjwatson: but bdmurray said you might be able to do something with the logs that are already attached
[06:44] <evand> cjwatson: any chance of getting another cd out today since linux-meta finished building?
[06:51] <cjwatson> mojonixon: you could run 'anna-install strace-udeb' on tty2
[06:51] <cjwatson> it was in universe in dapper, though
[06:51] <cjwatson> what I always used to do was 'udpkg -i /cdrom/pool/main/s/strace/strace_*.deb'
[06:51] <cjwatson> (which is horrible, but works)
[06:52] <cjwatson> blackskad: yeah, it was really a kernel bug but we already got a slew of bugs crashing in zlib_inflate and I think we fixed that
[06:53] <cjwatson> it's probably dependent on CD read errors, hence the poor reproducibility
[06:53] <cjwatson> feel free to reject that one
[06:53] <blackskad> cjwatson: ok, thanks :)
[06:53] <cjwatson> evand: sure, one moment
[06:53] <evand> thanks
[06:53] <cjwatson> I was busy doing interviews
[06:54] <cjwatson> hmm, linux-meta doesn't seem to have been accepted on all architectures
[06:54] <mojonixon> got it... thx.
[06:54] <evand> whoops
[06:56] <cjwatson> evand: never mind, my mistake, it's there
[06:56] <cjwatson> CDs building
[06:56] <evand> great
[07:09] <evand> hrm, what exactly is the standard procedure for translated strings conflicts in merges?
[07:10] <cjwatson> run in circles, scream and shout
[07:10] <cjwatson> the Ubuntu changes are typically Ubuntu branding of some sort
[07:10] <cjwatson> so s/Debian/Ubuntu/ - normally, I take the new Debian side and re-brand it
[07:11] <cjwatson> unfortunately some languages transliterate, decline, or otherwise mangle the word "Debian"
[07:11] <cjwatson> https://wiki.ubuntu.com/DistributionDefaultsAndBranding records (down at the end) some of the lore I've picked up about this
[07:11] <evand> haha
[07:12] <evand> thanks
[07:12] <cjwatson> after a bzr merge conflict, .OTHER is the Debian side
[07:13] <cjwatson> if you use vim, I have some vim macros to help
[07:13] <cjwatson> map ,u RUbuntu<Esc>
[07:13] <cjwatson> map ,z :let savelast=@/<CR>j?^msgid<CR>:let @/=savelast<CR>O#, fuzzy<Esc>
[07:13] <cjwatson> ,u just replaces the next six characters with Ubuntu (a surprising time-saver), while ,z marks the current string as fuzzy in case you don't know how to deal with the branding (that's the case for a lot of Arabic and Indic languages)
[07:14] <evand> nice!
[07:14] <evand> any particular reason why you don't maintain (or haven't published) kbd-chooser in bzr?
[07:15] <cjwatson> it's obsoleted by console-setup, so I just haven't bothered to import it
[07:15] <cjwatson> I basically just ignore it :)
[07:16] <evand> heh
[07:17] <cjwatson> one of these days I will either do something with the merge or remove the package entirely, but it's not worth much attentioon
[07:17] <cjwatson> -o
[07:18] <cjwatson> evand: new alternate dailies built
[07:18] <evand> great!
[08:13] <evand> heh, I probably should've read this first: http://cdimages.ubuntu.com/daily/20070530.2/report.html
[09:20] <cjwatson> actually neither of those should be a serious problem
[09:20] <cjwatson> (also, cdimages.u.c is deprecated, cdimage. is preferred)
[09:20] <cjwatson> -amd64-generic is a transitional metapackage, and gnome-orca is optional
[09:23] <evand> ah, I'll remember that for future reference.  It dies on gnome-orca.
[09:27] <evand> To clarify: gnome-orca depends on libgnome-speech3 but is not installable -> broken packages -> aptitude failed (100)
[10:52] <cjwatson> evand: mm, odd that it's being installed unless you asked for accessibility though
[10:55] <evand> I don't believe I did
[11:04] <evand> hrmm, it might be working this time around.  Odd.
[11:04] <evand> I stand corrected
[11:05] <evand> same problem
[11:05] <cjwatson> clearly gnome-orca ought to be fixed either way
[11:06] <cjwatson> evand: does the newt interface seem unusually laggy to you?
[11:06] <cjwatson> I'm wondering whether it's just this vmware run
[11:06] <evand> no, quite quick actually
[11:06] <cjwatson> odd
[11:06] <cjwatson> it seems to lag in responding to keystrokes
[11:06] <cjwatson> relative to feisty, I mean
[11:07] <cjwatson> oh, gnome-orca is a recommends of desktop
[11:07] <cjwatson> my bad
[11:08] <evand> I blame vmware
[11:08] <evand> about the lag, not the dependencies
[11:09] <cjwatson> gnome-orca has weird hardcoded dependency lines
[11:10] <cjwatson> I'm going to run away from this and ask dholbach about it next time I see him
[11:10] <evand> haha, ok
[11:10] <evand> had a chance to look over that email?
[11:20] <cjwatson> evand: mailed you back now
[11:23] <evand> thanks!