[15:14] <ev> cjwatson: might I trouble you for your opinion on whether this is the approach you'd take: http://paste.ubuntu.com/573484/ http://paste.ubuntu.com/573486/ ?
[15:14] <ev> I suppose we could alternatively set a debconf question, but I'm worried about maintaining the state of that when things go wrong.
[15:17] <cjwatson> yeah, that seems OK to me
[15:17] <cjwatson> debconf questions aren't really good for per-partition state
[15:17] <ev> indeed
[15:17] <ev> awesome, thanks a bunch
[15:21] <cjwatson> is this related to bug 726131, btw?
[15:21] <ubot2> Launchpad bug 726131 in debian-installer "alternate ISO: installation freezes when starting partman" [Critical,Confirmed] https://launchpad.net/bugs/726131
[15:23] <cjwatson> I haven't been able to reproduce that
[15:29] <ev> possibly. I'm trying to track it down, but my work environment is a bit of a mess, due to the partman-auto work.
[15:31] <ev> I've noticed we're not handling PartedServerError exceptions particularly well at the moment.  I should probably fix that along the way.
[15:31] <cjwatson> ev: I guess you've seen the comments on bug 722198?  looks like that bug should be reopened maybe
[15:31] <ubot2> Launchpad bug 722198 in oem-priority "installation hangs on 15reuse w/ blank disk" [Critical,Confirmed] https://launchpad.net/bugs/722198
[15:31] <CIA-3> partman-auto: evand * r594 ubuntu/ (4 files in 3 dirs):
[15:31] <CIA-3> partman-auto: Signal to clear_partitions in partman-target that it should not
[15:31] <CIA-3> partman-auto: notify the user about the unformatted partitions we have set up.
[15:32] <ev> indeed, I've added a comment asking for a ps tree
[15:32] <ev> just to get a better handle on what's hanging
[15:37] <CIA-3> partman-target: evand * r914 ubuntu/ (check.d/system_partitions_formatted debian/changelog):
[15:37] <CIA-3> partman-target: Filter out partitions that are being used by the reuse
[15:37] <CIA-3> partman-target: autopartitioning recipe.
[16:18] <CIA-3> partman-auto: cjwatson * r595 ubuntu/ (3 files in 3 dirs):
[16:18] <CIA-3> partman-auto: Guard against 'grep -s DISTRIB_ID $mountpoint/etc/lsb-release' returning
[16:18] <CIA-3> partman-auto: false and causing parted_server communications to get out of sync
[16:18] <CIA-3> partman-auto: (LP: #726131).
[16:18] <cjwatson> ev: ^- probably fixes a bunch of this kind of bug
[16:20] <ev> ah, yikes
[16:20] <ev> sorry about that
[16:20] <ev> thanks for the quick fix as well
[16:21] <cjwatson> one of those errors whose symptoms are distant from its cause :-/
[16:21] <cjwatson> anything more for partman-auto or should we upload this?
[16:22] <ev> by all means
[16:24] <CIA-3> partman-auto: cjwatson * r596 ubuntu/debian/changelog: releasing version 93ubuntu8
[16:37] <CIA-3> partman-target: evand * r915 ubuntu/debian/changelog: releasing version 71ubuntu2
[18:02] <ev> ugh, I'd love to know why the reuse autopartition recipe ends up in partman/choose_partition
[18:02]  * ev digs further
[18:02] <cjwatson> where's the source file?
[18:09] <ev> automatically_partition/reuse/{choices,do_option}
[18:09] <ev> cjwatson: unless I'm misunderstanding you.
[18:10] <ev> oddly it works in d-i the first time around
[18:10] <ev> subsequent attempts don't
[18:10] <CIA-3> debian-installer: cjwatson * r1418 ubuntu/ (build/pkg-lists/base debian/changelog): Include biosdevname-udeb on architectures where it's available.
[18:11] <cjwatson> I think I may be misunderstanding you - what do you mean by "ends up in partman/choose_partition"?
[18:13] <stgraber> hmm, before I go dig in debconf, anyone knows why on natty I'm being prompted by d-i for keyboard layout information when console-setup/layoutcode is set in the kernel boot options ?
[18:14] <cjwatson> try keyboard-configuration/layoutcode instead
[18:14] <cjwatson> (still a bug though, it should handle either for migration)
[18:14] <stgraber> oh, changed name ? :) ok, will preseed both then
[18:15] <cjwatson> yes, I still haven't jumped up and down on Anton's head as much as I should about that
[18:15] <stgraber> hehe
[18:16] <ev> cjwatson: if I try the reuse option, I get to the partman confirmation dialog (in d-i), if I hit back, go through the autopartition options again selecting reuse once more, it gets wedged and asks partman/choose_partition along the way.
[18:16] <stgraber> hmm, now it's stuck in the d-i menu with "Configure the keyboard" selected. Pressing enter doesn't do anything (menu disappears and re-appears after a second)
[18:17] <cjwatson> ev: hm, I recommend DEBCONF_DEBUG=developer + set -x in /lib/partman/lib/base.sh
[18:17] <cjwatson> (DEBCONF_DEBUG=developer as a boot param, I mean)
[18:17] <cjwatson> stgraber: if you get me a case I can reproduce quickly in kvm, then I can have a look
[18:18] <stgraber> cjwatson: I'm simply booting with: preseed/url=http://www.stgraber.org/download/preseeds/vmmanager-edubuntu.seed debian-installer/locale=en_US.UTF-8 keyboard-configuration/layoutcode=us netcfg/get_hostname=edubuntu-natty01 netcfg/disable_dhcp=true netcfg/get_ipaddress=172.16.3.254 netcfg/get_netmask=255.255.255.0 netcfg/get_gateway=172.16.3.1 netcfg/confirm_static=true netcfg/get_nameservers=8.8.8.8 netcfg/get_domain=ubuntu.com
[18:18] <stgraber> you shouldn't even need all the network stuff as it doesn't get that far
[18:19] <ev> cjwatson: http://paste.ubuntu.com/573576/ - is probably a better explanation, but indeed. I've been carefully placing set -x, will keep at it.
[18:20] <ev> (line 4056 is where I backed up to the menu)
[18:21] <cjwatson> ev: not sure of the exact sequence (dinnertime here), but it may be normal to ask partman/choose_partition
[18:21] <ev> oh
[18:21] <cjwatson> we sort of assume that if you back up from the confirmation question then you probably want more control
[18:21] <cjwatson> if it hangs then that isn't normal of course
[18:22] <ev> it doesn't hang, it just ends up with this http://people.canonical.com/~evand/tmp/madness.png
[18:22] <ev> but noted
[18:22] <cjwatson> that looks like the debconf protocol being out of sync, maybe
[18:23] <ev> hm
[18:23] <cjwatson> to debug that sort of thing I normally walk through DEBCONF_DEBUG=developer output from the last point at which it looked normal and verify sanity of the question/answer sequence
[18:23] <cjwatson> it can happen if non-debconf noise lands on stdout
[18:23] <cjwatson> though there are a couple of other possible causes
[18:24] <ev> ah, okay
[18:24] <ev> thanks
[18:24] <ev> will do
[18:26] <cjwatson> stgraber: reproduced, will fix later tonight
[18:26] <stgraber> cjwatson: ok, thanks
[19:21] <CIA-3> console-setup: cjwatson * r386 ubuntu/debian/ (3 files):
[19:21] <CIA-3> console-setup: Move keyboard detection templates from console-setup.templates to
[19:21] <CIA-3> console-setup: keyboard-configuration.templates.
[19:24] <CIA-3> console-setup: cjwatson * r387 ubuntu/debian/changelog: releasing version 1.57ubuntu8
[19:33] <CIA-3> ubiquity: cjwatson * r4522 trunk/ (d-i/manifest debian/changelog):
[19:33] <CIA-3> ubiquity: Automatic update of included source packages: base-installer
[19:33] <CIA-3> ubiquity: 1.116ubuntu1, console-setup 1.57ubuntu8, flash-kernel 2.28ubuntu15,
[19:33] <CIA-3> ubiquity: partman-auto 93ubuntu8, partman-target 71ubuntu2.
[19:44] <CIA-3> ubiquity: cjwatson * r4523 trunk/ (debian/changelog src/panel/panel.c):
[19:44] <CIA-3> ubiquity: Revert "Queue a redraw of the panel after setting the background". This
[19:44] <CIA-3> ubiquity: change is implicated in a memory leak leading to OOM conditions and
[19:44] <CIA-3> ubiquity: eventual crash of Xserver. (Possible fix for LP: #714829, would reopen
[19:44] <CIA-3> ubiquity: 693300)
[19:53] <CIA-3> ubiquity: cjwatson * r4524 trunk/ (debian/changelog src/panel/panel.c): Fix reference leak in panel set_background function.
[19:59] <CIA-3> ubiquity: cjwatson * r4525 trunk/debian/changelog: releasing version 2.5.19
[20:13] <brendand> anyone happen to know if the maverick desktop installer tries to configure DHCP at any point (maybe in the background?) like alternate?
[20:20] <cjwatson> brendand: we run nm-applet in the installer session, and that will normally try to bring up the network
[20:21] <brendand> cjwatson - my reference for the next question is another OS, but is it the case that if the ethernet link light isn't on then a driver hasn't been loaded?
[20:22] <cjwatson> I'm afraid I don't know offhand
[20:22] <brendand> cjwatson - cutting to the chase, we have a couple of systems that barf on DHCP config with alternate, but not desktop
[20:22] <brendand> cjwatson - the difference is no port activity in alternate
[20:22] <cjwatson> well, they are using entirely different software stacks ...
[20:23] <cjwatson> wireless encryption?
[20:23] <brendand> cjwatson - desktops. i assume they only have eth0 to go on
[20:23] <cjwatson> most likely then the driver is missing from d-i
[20:24] <cjwatson> file a bug on debian-installer in Ubuntu with details of what driver is used to drive the network card
[20:24] <cjwatson> it'll probably actually need to be fixed by the kernel guys
[20:31] <brendand> cjwatson - oh, by the way. was a bit of misinformation in my original question
[20:31] <brendand> cjwatson - it was *lucid* that is giving the problem
[20:32] <brendand> cjwatson - maverick is actually working fine
[20:32] <brendand> cjwatson - drivers missing make a lot of sense now
[20:32] <brendand> cjwatson - bugs shall be raised...
[20:33] <cjwatson> ah, yeah, this is the sort of thing we often fix from release to release
[20:33] <cjwatson> I imagine it would be easily backportable for 10.04.3
[21:00] <ev> ah ha!
[21:00] <ev> detected_filesystem doesn't always exist
[21:04] <ev> which is a bit strange
[21:22] <CIA-3> partman-auto: evand * r597 ubuntu/ (automatically_partition/reuse/do_option debian/changelog): Handle detected_filesystems not existing.
[21:32] <CIA-3> debian-installer: cjwatson * r1419 ubuntu/debian/changelog: releasing version 20101020ubuntu19