[15:51] <kirkland> cjwatson: can you point me to the relevant code for adding netboot users, such that I can fix https://bugs.edge.launchpad.net/ecryptfs/+bug/317895
[17:16] <cjwatson> kirkland: the netboot installer is just d-i, so user-setup
[17:23] <mark> hello
[17:23] <mark> can the ubuntu installer do fully automatic preseeding of an install with software raid and lvm?
[17:23] <mark> in hardy
[17:24] <cjwatson> not raid in hardy I'm afraid, though lvm should be possible
[17:25] <cjwatson> raid is possible post-hardy although not yet terribly well integrated in some ways
[17:25] <mark> thanks
[17:26] <mark> too bad, although we also do non-LTS releases on servers, it's typically those servers with software raid/lvm that need hardy
[17:26] <cjwatson> yeah, sorry about that
[17:27] <mark> oh well, life will get better in 2010! ;)
[17:33] <cjwatson> I know our security team are looking forward to having three LTS releases to support simultaneously ;-)
[17:34] <cjwatson> you probably *can* preseed RAID setup by hand in hardy (i.e. preseed/early_command that writes out /lib/partman/display.d/ script that calls mdadm lots), but it wouldn't be very much fun
[17:35] <mark> hmm
[17:35] <cjwatson> on the general principle that you can do anything with preseeding ...
[17:35] <mark> software raid with lvm setups are a bit annoying to do manually as well
[17:35] <mark> since you cannot do partitions in a raid volume in the installer
[17:36] <mark> and have to setup a separate raid volume for root, since grub can't boot from lvm either
[17:36] <cjwatson> right, also because there's no group selection
[17:36]  * cjwatson wonders if he should hijack the d-i team at debcamp with a whiteboard and redesign the whole thing
[17:37] <mark> that would be nice :)
[18:41] <charlie-tca> Ubiquity is broken on todays desktop cd?
[18:53] <mark> hrm, the installer doesn't see partitions inside sw raid arrays at all, not even if I make them on the command line
[18:54] <mark> oh well
[19:42] <mark> argh
[19:42] <mark> so for grub, having /boot on a separate non-LVM partition is not enough, it needs / as well?
[19:42] <mark> there really is no nice way out of this is there :)
[20:42] <cjwatson> charlie-tca: not that I'd heard - details please?
[20:42] <cjwatson> mark: it should see them provided that you set them up before the partitioner starts
[20:43] <mark> I did
[20:43] <mark> it did see the raid array, but not its partitions
[20:43] <mark> or does it cache that info, such that a second run won't work?
[20:43] <cjwatson> yes, it caches it
[20:43] <cjwatson> rm -rf /var/lib/partman and then rerun would do it
[20:43] <charlie-tca> cjwatson: Non of the Ubuntu or Xubuntu LiveCD's will install today. Ubiquity crashes. bug 322978
[20:43] <cjwatson> (nuke all partitioner state)
[20:44] <cjwatson> ...
[20:44] <charlie-tca> Well bug 332978
[20:44] <cjwatson> I can't see that either
[20:44] <charlie-tca> I will make it public
[20:44] <cjwatson> is it a user-setup crash? that one *is* known
[20:45] <charlie-tca> now it is visible
[20:45] <cjwatson> numpy, interesting
[20:45] <cjwatson> hadn't heard of this one
[20:45] <charlie-tca> ubiquity crashed with ImportError in <module>()
[20:45] <cjwatson> yes yes I'm already past that :)
[20:46] <cjwatson> ok, straightforward missing dep by the looks of things
[20:46] <cjwatson> charlie-tca: try installing python-numpy and then trying again?
[20:46] <charlie-tca> okay. I'll let you know
[20:48] <cjwatson> I don't get why Evan used a full-blown array here
[20:51] <cjwatson> charlie-tca: thanks for spotting that; I'm not sure I want to just add the dependency though, as it seems strange to me that this particular data type is necessary
[20:52] <cjwatson> and python-numpy is 1.5MB of .deb so I don't want to pull it onto the live CD unless it's strictly necessary
[20:52] <charlie-tca> I just tried to do the install. I don't know what really happened.
[20:52] <charlie-tca> I seem to know enough to tell you
[20:52] <cjwatson> right, I'm just explaining why it might take until tomorrow daytime before we fix it
[20:52] <charlie-tca> Thanks
[21:02] <charlie-tca> cjwatson: that does allow ubiquity to start. But it pulled in 'libblas3gf', 'libgfortran3', and 'libapack3gf' also.
[21:44] <cjwatson> charlie-tca: indeed, hence my comment about it being a heavy dependency
[21:44] <charlie-tca> Yeah, that makes sense. Unfortunately, there is more wrong. It won't go past the partitioner now.
[21:45] <charlie-tca> I'm thinking I'll wait and try this again tommorrow or next day.
[21:46] <cjwatson> between now and tomorrow, you have a whole lot of the installer team sleeping; next day is probably more realistic for an actual fix :)
[21:46] <cjwatson> (the build happens quite erly in the morning)
[21:46] <cjwatson> there are a few possibly-relevant fixes in bzr
[21:50] <charlie-tca> That works for me. At least we know it is broken, huh.
[21:51] <charlie-tca> Thanks for your time.
[22:29] <CIA-3> installation-guide: cjwatson * r448 ubuntu/debian/changelog: user-setup encryption change fixes LP: #323174
[23:03]  * mpt compares <http://upload.wikimedia.org/wikipedia/commons/5/51/Usb-creator.png> vs. <http://imag.malavida.com/mvimgbig/download/fedora-live-usb-creator-5419-1.jpg>
[23:04] <mpt> Ours is more explanatory
[23:04] <mpt> Fedora's has the advantage that it runs from Windows :-)