[04:19] <twb> Lucid's console-setup is calling "setupcon --force" in its postinst, which is screwing me over.  I'd like to find out why this change was introduced before filing a bug report; the Maintainer is listed as the install team.
[04:30] <persia> twb: I can't say for sure *why* that was introduced, but it appears to have been introduced as part of handling bug #524439
[04:31] <persia> There's some lengthy expositions in that bug that talk about how console-setup is to work with upstart and plymouth, which may help explain
[04:34] <persia> twb: From a quick look at postinst, it appears that it only calls `setupcon --force -k`under some conditions : I'm not sure precisely what you are doing that causes you to experience an issue, but perhaps you can cause an alternate activity to happen based on the right preconditions.
[04:35] <persia> (note that this may be all completely incorrect : I'm basing the pointers off the results of `bzr annotate` which may miss cases where code is moved en-masse from one place to another)
[04:46] <twb> persia: I tried forcing plymouth to be installed, but deboostrap installs console-setup as part of the base BEFORE any of my extra packages are installed
[04:47] <twb> Basically I'm building Lucid netboot images for a customer, from my Debian Sid host.
[04:48] <twb> If I run live-helper on the tty, setupcon runs inside the chroot and just breaks my font.  But if I do it over ssh, it turns off the local head, ignores ctrl-alt-del, breaks acpid's power button support, and kills all my ssh sessions -- basically meaning I have to do a hard reboot.
[04:48] <twb> If I boot the host OS with video=normal (disabling fbcon), the problem doesn't manifest.
[04:49] <persia> Someone else with deeper knowledge may be able to help.  Last traffic I saw was that live-helper was specifically unsupported, but that doesn't mean you shouldn't be able to make it work with a bit of fiddling.
[04:50] <twb> I could probably reproduce this with pure debootstrap
[04:51] <twb> I figured that "we don't support lh" just meant that Ubuntu wasn't supported as a "host"; not a "guest".
[04:52] <persia> I don't know enough about lh to know.
[04:52] <twb> persia: I did investigate how Ubuntu built its live CDs, but it turned out it was using lots of internal code that "won't work for anyone else without a lot of work"
[04:52] <persia> And I don't mean to imply that just because lh is involved that the issue doesn't matter, only to provide warning of the edges.
[04:53] <twb> Nod.
[04:53] <twb> persia: there's a reason you and cjwatson are in my IRC client's "bright" list :-)
[04:54] <persia> I've gotten livecd-rootfs to work locally without much extra work, but I've always just used that output to modify existing images (perhaps also modifying the kernel and initramfs).
[04:55] <persia> So *modifying* an Ubuntu CD isn't that hard: it's the creation of one that seems to be complicated.
[04:55] <persia> (I know folk who have gotten debian-cd running locally, but apparently it's very tricky)
[08:03] <cjwatson> twb: *blink* if it's breaking something, that's unambiguously a bug, but it's mostly been like that for years; the only recent change to that part of the code was meant to be to cope with plymouth.
[08:04] <cjwatson> twb: which branch in the postinst is it taking?
[08:04] <twb> I'm not 100% sure
[08:05] <twb> It only happens if I boot the host sid system with video=i915:modeset=1
[08:05] <twb> I'm avoiding it right now by booting with modeset=0, so I guess I'm still using the fbcon, even
[08:06] <twb> I suppose I could strace -f the whole build to find out how it's invoking setupcon
[08:12] <ev> I started playing around with adding unit tests to ubiquity yesterday but ran into the wall of not being able to pass PATH through sudo.
[08:13] <twb> ev: can't you instruct sudoers to allow it through?
[08:13] <twb> Or even just sudo foo --> sudo env "PATH=$PATH" foo
[08:15] <ev> I'd like to avoid having to do that, if possible
[08:15] <ev> and the env trick doesn't seem to work either
[08:15] <twb> Fair enough
[08:21] <ev> ah, -i
[08:21] <ev> thanks twb
[08:22] <twb> Oh, sorry, I misunderstood.  I thought you were using -i, but *wanted* to preserve the PATH.
[08:25] <ev> yeah, ignore me.  It works fine without -i
[08:25] <ev> but it looks like I'm going to need to run the tests in a chroot anyway
[08:27] <twb> Usually, -i is more Right than -H -s or "sudo su -".
[10:00] <CIA-3> grub-installer: cjwatson * r843 ubuntu/ (debian/changelog grub-installer): Fix GPT detection to account for changed fdisk output (LP: #538100).
[10:02] <CIA-3> grub-installer: cjwatson * r844 ubuntu/debian/changelog: releasing version 1.49ubuntu7
[10:46] <CIA-3> partman-base: cjwatson * r191 ubuntu/ (debian/changelog init.d/parted):
[10:46] <CIA-3> partman-base: Don't source /usr/share/debconf/confmodule in init.d/parted;
[10:46] <CIA-3> partman-base: /lib/partman/lib/base.sh will always source it for us.
[11:21] <CIA-3> usb-creator: evand * r275 usb-creator/ (debian/changelog usbcreator/backends/udisks/backend.py):
[11:21] <CIA-3> usb-creator: Rework partition mounting so that it doesn't fail if the partition
[11:21] <CIA-3> usb-creator: was mounted between dbus calls.
[11:46] <CIA-3> console-setup: cjwatson * r138 ubuntu/ (console-setup-tty debian/changelog): console-setup-tty: Only call setfont once, for both font and ACM.
[13:22] <CIA-3> ubiquity: cjwatson * r3921 ubiquity/ (bin/ubiquity-wrapper debian/changelog): Explicitly pass $HOME through kdesudo (LP: #538142).
[14:25] <CIA-3> ubiquity: cjwatson * r3923 ubiquity/debian/changelog: categorisation
[14:30] <CIA-3> ubiquity: cjwatson * r3924 ubiquity/ (bin/ubiquity-wrapper debian/changelog debian/control):
[14:30] <CIA-3> ubiquity: Use manual xauth code and sudo rather than kdesudo; the latter does
[14:30] <CIA-3> ubiquity: not permit passing through environment variables other than
[14:30] <CIA-3> ubiquity: one-by-one, which broke various things (LP: #526456, #538142).
[14:37] <CIA-3> partman-target: cjwatson * r789 ubuntu/ (debian/changelog finish.d/create_fstab_header): Mount /proc with nodev,noexec,nosuid.
[14:39] <CIA-3> partman-target: cjwatson * r790 ubuntu/debian/changelog: releasing version 64ubuntu7
[14:39] <CIA-3> ubiquity: evand * r3925 ubiquity/ (debian/changelog gui/gtk/stepLanguage.ui): Align labels to the top on the language page (LP: #532876).
[14:45] <CIA-3> ubiquity: cjwatson * r3926 ubiquity/ (debian/changelog ubiquity/components/ubi-usersetup.py):
[14:45] <CIA-3> ubiquity: Construct the default hostname from the text in the username box, not
[14:45] <CIA-3> ubiquity: from the previous text in the hostname box (LP: #527704).
[15:04] <twb> In Lucid, I need casper to be in rcS, but *not* be in rc0.
[15:04] <twb> "update-rc.d casper disable 0" doesn't DTRT
[15:04] <twb> Oops, wrong channel.
[15:12] <cjwatson> ev: have you heard anything about the Kashmir bug yet?
[15:14] <ev> cjwatson: I have an exchange going with Ken and company about it
[15:14] <ev> I can CC you in if you'd like
[15:15] <cjwatson> no that's ok, I just wanted to know it was moving
[15:15] <cjwatson> there wasn't anything on the bug so I wasn't sure
[15:15] <ev> okay
[15:15] <ev> there seems to be some confusion as to who on the design team is tackling it, but I'm working on that.
[15:19] <cody-somerville> When you boot with casper, should the live user it creates show up in /etc/passwd in the live system?
[15:20] <cjwatson> I'd expect so
[15:20] <cjwatson> it creates it with adduser
[15:31] <cody-somerville> cjwatson, What would cause casper to fail to create the user?
[15:36]  * ev HATES PYTHON DEBUGGERS
[15:36] <ev> *ahem*
[15:37] <cjwatson> cody-somerville: don't know, check /var/log/casper.log
[15:38] <twb> ev: gimme slime any day, eh?
[15:38] <twb> cody-somerville: maybe combined with break=casper-bottom
[15:39] <ev> twb: I'd be happy with gdb.  At least it has features.
[15:41] <cody-somerville> There are several 'invalid user ubuntu' messages after "Begin: Adding live session user...\nsed: bad option in substitution expression"
[15:44] <twb> cody-somerville: what version of casper?
[15:45] <cody-somerville> 1.228
[16:15] <cjwatson> cody-somerville: I'm not seeing anything obvious - but if you add the 'debug' boot option then you'll get a full 'set -x' shell trace in /dev/.initramfs/initramfs.debug, which you can correlate against the code to work out what's going wrong
[16:15] <cody-somerville> cjwatson, k, thanks.
[16:16] <cjwatson> cody-somerville: DEBCONF_DEBUG=developer mightn't hurt either
[16:25] <CIA-3> ubiquity: evand * r3927 ubiquity/ (3 files in 3 dirs):
[16:25] <CIA-3> ubiquity: The run method of GTK+ dialogs enters a nested main loop. However,
[16:25] <CIA-3> ubiquity: the user can click the quit button (which calls the quit dialog's
[16:25] <CIA-3> ubiquity: run method) before the page is ready. Once the page is actually ready,
[16:25] <CIA-3> ubiquity: another main loop is entered, and thus if the user hits the quit button
[16:25] <CIA-3> ubiquity: on the quit dialog at this point, nothing will happen. Work around this
[16:25] <CIA-3> ubiquity: by managing the state of the quit dialog in the frontend.
[16:36] <kirkland> is "installing the base system" really crawling for anyone else today?
[16:38] <davmor2> kirkland: is this in vm
[16:38] <kirkland> davmor2: no, real hardware, install the server
[16:39] <davmor2> 32 or 64 bit?
[16:39] <cjwatson> recent change to dpkg to improve reliability had a performance hit
[16:39] <cjwatson> it's being discussed on debian-dpkg
[16:39] <kirkland> 64bit
[16:40] <cjwatson> shouldn't matter
[16:41] <kirkland> cjwatson: what was under a 10 minute install is ~30 minutes and counting now
[16:42] <cjwatson> yes I know
[16:43] <cjwatson> I'm sorry but we're looking into it.
[16:43] <cjwatson> the bug that was closed as part of this had 200 duplicates, so it isn't as simple as just reverting
[16:43] <cjwatson> as a workaround, install using ext3
[16:44] <kirkland> cjwatson: okay, thanks
[16:46] <kirkland> cjwatson: what's the d-i partman/foo string ext4 option I'm looking for to add to my preseed to go ext3 rather than ext4?
[16:47] <cjwatson> https://help.ubuntu.com/9.10/installation-guide/i386/preseed-contents.html#preseed-partman  search for "default filesystem"
[16:48] <twb> Is Ubuntu waiting for Debian for btrfs/aufs/squashfs udebs for d-i?
[16:49] <davmor2> cjwatson: I just got a the kernel failed to re-read the partitions
[16:54] <davmor2> kirkland: yes does seem slower, but I don't think it's going as slow as yours was
[16:55] <kirkland> cjwatson: perfect, thanks
[16:55] <kirkland> davmor2: vm, or hw?
[16:55] <davmor2> hw
[17:00] <davmor2> kirkland: nearly done, but you're right it is slower
[17:02] <kirkland> cjwatson: ext3 *much* faster, install finished in ~6 minutes
[17:03] <twb> kirkland: compared to ext4?
[17:03] <kirkland> twb: working around a dpkg bug on ext4
[17:03] <twb> Righto.
[17:04] <twb> Otherwise I was gonna wave -ocommit and SSD erase block alignment and such at you
[17:09] <davmor2> kirkland: I see what you mean now linux-headers-2.6.32-16-server is mega slow been on this now for serveral minutes
[17:10] <kirkland> davmor2: yup
[17:10] <kirkland> davmor2: i killed the install after ~35 minutes on a dual core laptop, 2.4GHz, 4GB, 7200rpm disk
[17:11] <davmor2> it's carried on now
[17:11] <davmor2> kirkland: so still no as slow as yours sorry ;)
[17:18] <cjwatson> davmor2: on what configuration?
[17:18] <cjwatson> davmor2: (please be verbose up-front!)
[17:20] <davmor2> cjwatson: iso is 0fb898975b9ab8acd21bc25ca19fca9733d007a17144ae5f75893aaa1a878768 *lucid-server-amd64.iso  .  I did the standard whole drive lvm install that is auto selected
[17:21] <cjwatson> lvm.  thank you.  that was the magic word
[17:21] <cjwatson> 32/64-bitness is not usually relevant to this kind of thing
[17:21] <cjwatson> davmor2: did you have any partitions mounted by any manual means?
[17:21] <davmor2> after base the other packages installed at normal speed it was only the base that was slow.  So I installed additional ssh and that was quite quick
[17:22] <davmor2> cjwatson: nope
[17:22] <davmor2> there were partitions on there priory that were over written
[17:22] <cjwatson> normal speed> just different kinds of packages.  packages with lots of small files, notably linux-headers, suffer particularly badly
[17:23] <davmor2> cjwatson: hope that makes sense
[17:23] <cjwatson> davmor2: please don't drip-feed me this information, is there a bug with all of the information there at once?
[17:23] <cjwatson> (perhaps an older one I haven't got to yet ...)
[17:24] <davmor2> cjwatson: I'll have a look this is the first time I've seen this a) slowness and b) error message
[17:24] <cjwatson> hm, haven't you mentioned it before?
[17:24] <cjwatson> maybe not
[17:24] <cjwatson> don't worry about the slowness, it's already very well-understood
[17:25] <cjwatson> and I don't think it can be related
[17:25] <davmor2> I don't think I have reported it let me check though
[17:29] <davmor2> cjwatson: I can't find a d-i bug in my bug mails so I haven't reported one
[17:29] <cjwatson> ok
[17:30] <davmor2> cjwatson: Would you like me to report a new bug or see if I can find one?
[17:31] <cjwatson> report a new one please
[17:31] <davmor2> np's
[17:31] <cjwatson> you know what I think about piggybacking on existing bugs :)
[17:34] <davmor2> that the piggy's should be shot and turned into bacon?
[17:40] <davmor2> meh what's the crash database bug number please persia
[17:51] <CIA-3> usb-creator: evand * r276 usb-creator/ (debian/changelog usbcreator/frontends/gtk/frontend.py): Somewhat fix progress reporting.
[17:53] <ev> michaelforrest1: https://bugs.edge.launchpad.net/ubuntu/+source/usb-creator/+bug/539189
[18:01] <davmor2> cjwatson: https://bugs.launchpad.net/ubuntu/+bug/539194  hope that makes sense fsckin' apport's still broken
[18:02] <cjwatson> thanks
[18:03] <davmor2> tried to make the old and new as obvious as possible let me know if there is anything else
[18:03] <cjwatson> I can't find a matching trace in the log - did you quote the error message exactly?  I want to grep for it
[18:04] <cjwatson> ah, there we go, found it, you didn't quite :)
[18:04] <cjwatson> ok, will queue up ...
[18:33] <davmor2> I was close though right ;)
[18:35] <davmor2> cjwatson: so that's okay you don't need anything else?
[21:57] <ev> is there some trick to creating a debconf database from scratch?  DEBCONF_DEBUG=db is being incredibly unhelpful.
[21:58]  * ev digs through the source
[22:06] <cjwatson> ev: debconf-loadtemplate
[22:07] <ev> indeed, that blows up.  I'm trying roughly this: http://paste.ubuntu.com/395848/
[22:10] <ev> which fails with: "Configuration database "configdb" was not initialized."
[22:10] <ev> slowly trying to trace through that, but my perl is a bit rusty
[22:17] <cjwatson> looks like about the right place to start ...
[22:27]  * ev bangs head repeatedly on table
[22:28] <ev> s/Name: config/Name: configdb/
[22:28] <cjwatson> hah, I have never done that before, oh no
[22:28] <cjwatson> except for all those times
[22:29] <ev> lol
[22:30] <sbeattie> cjwatson: is there a known issue with the alt installer failing on setting up swap with full-disk encryption?
[22:31] <cjwatson> not one I know in those terms
[22:59] <sbeattie> (sorry, meant guided partitioning w/ lvm+encryption)
[23:02] <cjwatson> can you be a bit more verbose about "failing" please?
[23:04] <sbeattie> cjwatson: sure, bug 539324
[23:38] <CIA-3> ubiquity: cjwatson * r3929 ubiquity/ (d-i/manifest debian/changelog):
[23:38] <CIA-3> ubiquity: Automatic update of included source packages: console-setup
[23:38] <CIA-3> ubiquity: 1.34ubuntu11, grub-installer 1.49ubuntu7, partman-base 138ubuntu3,
[23:38] <CIA-3> ubiquity: partman-target 64ubuntu7.
[23:47] <CIA-3> ubiquity: cjwatson * r3930 ubiquity/debian/changelog: releasing version 2.2.0