CIA-61 | debian-installer: cjwatson * r1002 ubuntu/ (4 files in 2 dirs): Move mainline architectures to 2.6.28-1 kernels. | 11:55 |
---|---|---|
CIA-61 | debian-installer: cjwatson * r1003 ubuntu/debian/changelog: releasing version 20081029ubuntu3 | 13:35 |
_ruben | base-installer: warning: attempting to install to unclean target | 13:57 |
_ruben | i get that after specifying some packages with pkgsel/include | 13:57 |
cjwatson | pkgsel/include has nothing to do with it | 14:10 |
cjwatson | you're trying to install to an existing root filesystem without formatting it | 14:10 |
_ruben | cjwatson: well, its reproducable with a clean disk, and reducint the pkgsel/include line to one param tops, it does work even with a disk with prev install on it | 14:16 |
_ruben | so pkgsel/include string postfix works, but postfix vim does not | 14:17 |
_ruben | and it complains *after* installing base system, which kinda seems odd | 14:18 |
_ruben | hrm .. its something fishy that's going on .. now im getting that error with just a single pkgsel/include param but basically unclean disk, yet i've been doing tons of test installs this way | 14:22 |
_ruben | time to investigate deeper | 14:22 |
_ruben | hmm .. clean disk + several included packages = unclean target | 14:35 |
_ruben | lets see if hardy behaves the same | 14:36 |
_ruben | cjwatson: when doing a guided partition (full disk, no lvm), does it reformat? | 14:39 |
_ruben | judging from the warning, i guess it does | 14:40 |
_ruben | so its a bug somewhere either way i'd say | 14:40 |
_ruben | same on hardy :/ | 14:46 |
CIA-61 | debian-installer: cjwatson * r1004 ubuntu/ (build/pkg-lists/netboot/ia64.cfg debian/changelog): Make nic-usb-modules optional on ia64. | 14:57 |
cjwatson | _ruben: logs, please? | 14:57 |
cjwatson | pkgsel/include really has absolutely nothing to do with it. It isn't processed until after that warning is emitted | 14:57 |
cjwatson | any link there is essentially coincidental and a red herring | 14:57 |
kirkland | cjwatson: so the adduser patch for --encrypt-home is positively trivial, i think ... http://pastebin.ubuntu.com/77388/ | 14:58 |
kirkland | cjwatson: i managed to take care of all of the "work" in ecryptfs-setup-private | 14:58 |
kirkland | cjwatson: i'd like to get someone to review that script, end to end, if possible; i'll talk to doko and/or slangasek, at your discretion; i'll probably have kees/jdstrand look at it too | 14:59 |
kirkland | cjwatson: here's the magic step, which i figured out and implemented yesterday ... | 14:59 |
_ruben | cjwatson: well, i removed the pkgsel/include line from my preseed, didnt use a clean disk, and the install completed as expected | 14:59 |
_ruben | cjwatson: which log(s) would you like to see? | 14:59 |
kirkland | cjwatson: a random passphrase is generated, and temporarily written to a 400 file | 15:00 |
kirkland | cjwatson: then, i hacked pam_ecryptfs.so 's password change mechanism to detect the existence of that file | 15:00 |
cjwatson | kirkland: looks fine to me, perhaps also explicitly check whether ecryptfs-setup-private is installed and emit a good error message | 15:00 |
kirkland | cjwatson: sounds good ... is "Recommends" the appropriate level? since most of the rest of adduser would operate fine without ecryptfs-utils | 15:01 |
cjwatson | kirkland: and we'd need to change user-setup to apt-install ecryptfs-utils before calling adduser | 15:01 |
kirkland | cjwatson: and if found, it's encrypted and written back to disk, and the cleartext file is removed | 15:01 |
cjwatson | kirkland: Recommends would normally seem plausible, but adduser is in a very central position in the system where that doesn't always do what you expect; I'd like to reserve judgement on whether it should be Recommends or Suggests | 15:01 |
cjwatson | _ruben: the installer's /var/log/syslog from an *unsuccessful* installation, please | 15:02 |
kirkland | cjwatson: fine by me... i suppose the question was more, "It should not be a 'Depends', right?" :-) | 15:02 |
cjwatson | _ruben: perhaps you're getting the warning anyway but it only happens to actually cause a problem if you're installing certain packages at the pkgsel stage | 15:02 |
cjwatson | _ruben: nevertheless, once that warning is emitted you're in unsupported territory | 15:03 |
cjwatson | kirkland: right, I'd agree with that | 15:03 |
_ruben | cjwatson: ok | 15:03 |
kirkland | cjwatson: i'll file a bug with patch against adduser momentarily ... it should not but uploaded, though, until i get the enhanced ecryptfs-utils in the archive | 15:04 |
cjwatson | kirkland: I'm going on leave tomorrow for a couple of weeks, so I'd rather somebody with a bit more continuity dealt with the sponsorship | 15:04 |
kirkland | cjwatson: that's fine, i understand ;-) | 15:04 |
kirkland | cjwatson: i was just updating you, with some confidence, that I did about 9 hours of dev/testing on this while in the car yesterday :-) | 15:05 |
cjwatson | good :-) | 15:05 |
kirkland | cjwatson: it's actually working quite well; the adduser piece is trivial, and the meat of the ecryptfs-setup-private code is common between encrypting home and private | 15:07 |
cjwatson | kirkland: sorry, I never got round to setting up a time for briefing you on the server installer stuff for UDS, and now I'm coming down with a cold so would rather not talk on the phone too much | 17:32 |
cjwatson | kirkland: where are you feeling underprepared, and I can try to run through it here? | 17:32 |
kirkland | cjwatson: well, it's actually thanksgiving day here, and i'm with the inlaws, so today is really bad for me | 17:33 |
kirkland | cjwatson: i don't have any poignant weaknesses, that i know of, at the moment | 17:34 |
kirkland | cjwatson: .... well, that sounds bad .... :-) | 17:34 |
cjwatson | yeah, sorry, I did mean to do this yesterday | 17:34 |
kirkland | cjwatson: no worries ... i was in the car for 9 hours yesterday :-) | 17:34 |
kirkland | cjwatson: good, offline time to hack hard on the homedir stuff | 17:35 |
kirkland | cjwatson: i just stubbed in the detection of ecryptfs-setup-private, with error message | 17:35 |
kirkland | cjwatson: i'm going to upload a pair of packages to my ppa | 17:35 |
kirkland | cjwatson: oh, one thing I could use ... | 17:47 |
CIA-61 | hw-detect: cjwatson * r97 ubuntu/debian/changelog: releasing version 1.70ubuntu3 | 17:47 |
kirkland | cjwatson: a couple of pointers to the two packages and/or bzr branches where the adduser --encrypt-home would theorectically go | 17:48 |
cjwatson | kirkland: lp:~ubuntu-core-dev/user-setup/ubuntu, I'd expect | 17:48 |
kirkland | cjwatson: same for both text and gui installs? | 17:50 |
cjwatson | user-setup is one of the installer components used by ubiquity | 17:50 |
cjwatson | you'll need to stick a UI on top of it for ubiquity, though | 17:50 |
cjwatson | so lp:~ubuntu-installer/ubiquity/trunk for that | 17:50 |
cjwatson | do not edit the stuff in d-i/source/ directly, at least not for upload - you'll find a copy of the user-setup source package in there though | 17:51 |
kirkland | cjwatson: gotcha, that was the part i was looking for (poorly asked question) | 17:51 |
kirkland | cjwatson: i suppose i can get assistance from evand for that, right? | 17:51 |
cjwatson | doing the glue for ubiquity is a bit tricky - right, ask evand | 17:51 |
cjwatson | evand: are we at the point of having a working ubiquity yet? | 17:56 |
cjwatson | evand: you seem to have been making pretty good progress on that ... | 17:57 |
kirkland | cjwatson: once the ecryptfs-utils is done building in my PPA, you can give the encrypted home functionality a test drive if you install my adduser and ecryptfs-utils, and then "adduser --encrypt-home $user" | 18:10 |
* kirkland really has to run for family time now | 18:10 | |
CIA-61 | user-setup: cjwatson * r128 ubuntu/ (29 files in 4 dirs): merge from Debian 1.23 | 18:12 |
cjwatson | kirkland: not for at least two weeks :-) | 18:12 |
CIA-61 | user-setup: cjwatson * r129 ubuntu/debian/po/templates.pot: fix up to reduce spurious conflicts | 18:20 |
CIA-61 | user-setup: cjwatson * r130 ubuntu/debian/po/ (41 files): more conflict avoidance | 18:22 |
CIA-61 | user-setup: cjwatson * r131 ubuntu/debian/po/ (cy.po ga.po se.po sr.po): msgmerge new translations | 18:30 |
CIA-61 | user-setup: cjwatson * r132 ubuntu/debian/changelog: releasing version 1.23ubuntu1 | 18:44 |
CIA-61 | console-setup: cjwatson * r69 hardy-proposed/debian/ (changelog console-setup.initramfs-top): | 18:56 |
CIA-61 | console-setup: Create /dev/tty[1-6] in our initramfs script before trying to operate on | 18:56 |
CIA-61 | console-setup: them (LP: #130444). | 18:56 |
CIA-61 | console-setup: cjwatson * r70 hardy-proposed/debian/changelog: releasing version 1.21ubuntu9 | 19:02 |
CIA-61 | console-setup: cjwatson * r78 console-setup/debian/ (changelog console-setup.initramfs-top): | 19:10 |
CIA-61 | console-setup: Create /dev/tty[1-6] in our initramfs script before trying to operate on | 19:10 |
CIA-61 | console-setup: them (LP: #130444). | 19:10 |
CIA-61 | console-setup: cjwatson * r79 intrepid-proposed/debian/changelog: releasing version 1.25ubuntu4 | 19:10 |
mo | someone here familia with the alternate install partition manager and encrypted partitions? | 23:47 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!