2008/02/29/#ubuntu-installer.txt

[01:03] <CIA-24> ubiquity: superm1 * r2517 ubiquity/ (2 files in 2 dirs): update to newer base sql file
[03:54] <CIA-24> ubiquity: superm1 * r2518 ubiquity/ (3 files in 3 dirs): add PVR-350 output suppport
[06:26] <CIA-24> ubiquity: superm1 * r2519 ubiquity/ (debian/changelog ubiquity/components/mythbuntu_apply.py): remove only \selected\ themes
[06:39] <CIA-24> ubiquity: superm1 * r2520 ubiquity/ (debian/changelog scripts/mythbuntu/apply-type): update mythweb's installation script
[06:42] <stuuf> anyone know if the ubuntu installer includes an ntfs resizer?
[06:45] <superm1> there is utility that gets called to resize ntfs yes
[06:45] <superm1> it however is an external utility, not directly "part" of the installer
[06:57] <stuuf> but is it included on the cd?
[06:58] <evand> stuuf, ntfsresize.  Yes, it's on the CD.
[06:59] <stuuf> ok, i thought they would have added that by now
[06:59] <superm1> now look what you did, scared evan away :)
[06:59] <superm1> it's been on the cd for some time i thought actually
[06:59] <superm1> i remember helping a friend install either dapper or edgy and having it do resizing
[07:04] <stuuf> ah well i 'm usually a gentoo person but i figured i'd try ubuntu on a semi-abandoned laptop for boredom-related reasons
[07:05] <stuuf> haven't used it much since breezy or whatever was around in fall 2005
[07:05] <superm1> *lots* has changed since then :)
[07:38] <stuuf> i wish it would show some of the output from ntfsresize though, it's just saying "resizing partition 0%" for the last 20 minutes
=== cjwatson_ is now known as cjwatson
[09:32] <cjwatson> stuuf: bug 14100
[09:32] <ubotu> Launchpad bug 14100 in partman-partitioning "partman doesn't provide ntfsresize status details" [Medium,Confirmed] https://launchpad.net/bugs/14100
[09:32] <cjwatson> (sorry, I got disconnected and wasn't sure that made it last time)
[09:33] <stuuf> ah well it installed everything eventually
[10:36]  * cjwatson blinks. Am I hallucinating or is a feasible solution for bug 181296 a three-liner?
[10:36] <ubotu> Launchpad bug 181296 in partman-auto "partman preseeding should not require localized strings" [Medium,Confirmed] https://launchpad.net/bugs/181296
[11:53] <CIA-24> partman-base: cjwatson * r78 ubuntu/ (debian/changelog lib/base.sh):
[11:53] <CIA-24> partman-base: * Backport from trunk (LP: #181296):
[11:53] <CIA-24> partman-base:  - Support preseeding questions asked through ask_user using the name of
[11:53] <CIA-24> partman-base:  the plugin responsible for the answer you want (e.g.
[11:53] <CIA-24> partman-base:  partman-auto/init_automatically_partition=biggest_free).
[11:56] <CIA-24> partman-auto: cjwatson * r253 partman-auto/ (debian/changelog lib/recipes.sh):
[11:56] <CIA-24> partman-auto: * Backport from trunk (LP: #181296):
[11:56] <CIA-24> partman-auto:  - Support preseeding partman-auto/choose_recipe using the name of the
[11:56] <CIA-24> partman-auto:  recipe you want (e.g. "home"), as an alternative to the localised
[11:56] <CIA-24> partman-auto:  description.
[11:56] <xivulon> At the end of the installation, I see a shell message with something like "The system is going to reboot, please remove any CD and press enter"
[11:56] <xivulon> is it possible to avoid that?
[11:58] <cjwatson> xivulon: yes, put 'noprompt' on the kernel command line
[11:58] <cjwatson> xivulon: re amd64, there is no possibility of it changing now
[11:59] <xivulon> I already have d-i finish-install/reboot_in_progress note and d-i cdrom-detect/eject boolean false in preseed...
[11:59] <xivulon> Plus the change in initd/ubqiuity
[11:59] <cjwatson> it's unfortunate that casper has separate configuration here
[11:59] <xivulon> does the noprompt implies something else?
[11:59] <cjwatson> no
[12:00] <xivulon> will add noprompt then!
[12:03] <xivulon> cjwatson any chance of changing ISO names from amd64 to 64bits or something?
[12:03] <cjwatson> no
[12:03] <xivulon> :(
[12:04] <cjwatson> feel free to use more descriptive text wherever it is visible, but the codename is 100% fixed
[12:04] <xivulon> Wubi displays the filename when downloading, and many intel users (100% in fact) think we got it wrong when the see "amd"...
[12:05] <cjwatson> feel free to use more descriptive text
[12:05] <cjwatson> such as we use on the CD image website
[12:05] <xivulon> Will do so
[12:06] <xivulon> But I feel in the long run the text should be different
[12:06] <cjwatson> we will not change the codename
[12:07] <xivulon> tried at least...
[12:07] <cjwatson> I'm sorry, but architecture codenames are there to be fixed identifiers, not to be pretty for users
[12:08] <xivulon> I appreciate that
[12:09] <CIA-24> partman-base: cjwatson * r79 ubuntu/debian/changelog: releasing version 114ubuntu2
[12:13] <CIA-24> partman-auto: cjwatson * r254 partman-auto/debian/changelog: releasing version 73ubuntu3
[12:14] <CIA-24> installation-guide: cjwatson * r413 ubuntu/ (debian/changelog en/appendix/preseed.xml): * Document new simplified partman preseeding (LP: #181296).
[12:19] <CIA-24> installation-guide: cjwatson * r414 ubuntu/debian/changelog: releasing version 20080211ubuntu2
[15:09] <xivulon> evand have uploaded rev 443 for wubi
[15:09] <xivulon> should be a good one
[15:10] <evand> ok, thanks
[15:10] <evand> I wont be able to get to it until later today (sometime after 3-6p EST), though
[15:11] <xivulon> no problem at all
[15:11] <xivulon> when you have time can you have a look at 186117/186114?
[15:11] <xivulon> you mentioned you could remount ntfs, but I tried and it does not work for me
[15:12] <xivulon> it used to work in gutsy, I am not sure what is the reason for such regression
[15:13] <evand> I'll give 186117 further testing once my laptop is usable (around the same time as when I can review rev 443)
[15:14] <evand> xivulon, can you please provide a proper diff for bug 186114
[15:14] <ubotu> Launchpad bug 186114 in wubi "umountroot cannot handle fuse host mounts" [Medium,Confirmed] https://launchpad.net/bugs/186114
[15:14] <evand> preferably in diff -u format
[15:15] <xivulon> will do but will be away on the w/e
[15:15] <evand> ok
[15:15] <xivulon> anyay the chunk in there is simply supposed to replace remount_ro function in umountroot
[15:16] <xivulon> it won't work anyway if ntfs cannot be remount
[15:16] <evand> indeed, I saw that note.
[15:22] <xivulon> see last comment in 186117 for the command/error in my case
[15:23] <evand> indeed
[15:25] <cjwatson> evand: do your map fixes in bzr address bug 195159?
[15:25] <ubotu> Launchpad bug 195159 in ubiquity "The new time zone / city choosing is annoying" [Medium,Confirmed] https://launchpad.net/bugs/195159
[15:26] <evand> it should.  I've marked it as fix committed and will follow up on the bug when the new ubiquity is released.
[15:28] <cjwatson> ok
[15:28] <cjwatson> are you planning to do that today? there's a lot of good stuff in bzr
[15:29] <evand> sure, I'll take care of it in a few minutes
[15:42] <evand> uhm, odd.  Anyone else getting connection refused to bazaar.launchpad.net over SSH?
[15:43] <cjwatson> yes
[15:43] <evand> ok, good
[15:43] <evand> well, not good, but better than it just being me.
[15:47] <cjwatson> evand: seems to be working again now
[15:48] <evand> ah, fantastic.  Thank
[15:48] <evand> s
[15:52] <CIA-24> ubiquity: evand * r2521 ubiquity/debian/po/ (79 files): debconf-updatepo
[15:57] <CIA-24> ubiquity: evand * r2522 ubiquity/ (d-i/manifest debian/changelog):
[15:57] <CIA-24> ubiquity: * Automatic update of included source packages: apt-setup
[15:57] <CIA-24> ubiquity:  1:0.31ubuntu6, console-setup 1.21ubuntu3, debian-installer-utils
[15:57] <CIA-24> ubiquity:  1.50ubuntu2, partman-auto 73ubuntu3, partman-base 114ubuntu2, user-
[15:57] <CIA-24> ubiquity:  setup 1.16ubuntu4.
[16:19] <cjwatson> evand: #ubuntu-devel
[16:19] <evand> indeed
[16:21] <davmor2> guy's quick query is bug 188085 likely to get fixed before alpha 6?
[16:21] <ubotu> Launchpad bug 188085 in debian-installer "debian-installers encrypted erase disc cancel button should read Skip" [Undecided,New] https://launchpad.net/bugs/188085
[16:21] <cjwatson> no, I don't think it's urgent
[16:21] <cjwatson> nor am I even sure it should be fixed at all
[16:23] <cjwatson> I've added a comment
[16:25] <davmor2> right I thought that if you hit cancel it did just that? in the other sections.  Always seems to for me..
[16:25] <cjwatson> for example, cancelling netcfg's DHCP progress bar doesn't cancel the whole install
[16:25] <cjwatson> I can't think of any example in d-i where cancel means cancel the whole install
[16:26] <cjwatson> you can go back to the main menu and select "Abort the installation"
[16:28] <davmor2> I think the only issue I have with this section in particular is rather than it taking a few seconds of minutes it takes hours on a small drive yet alone a large drive.  and if your installing on a new drive you don't need to delete it anyway.
[16:33] <cjwatson> in any case, at present there is no (sane) way to override the text just for that instance
[16:34] <cjwatson> we could perhaps change the text to indicate that you can safely cancel
[16:35] <davmor2> cjwatson: that would be a big help.
[16:35] <cjwatson> have to do that in d-i upstream though
[16:35] <cjwatson> and I think they're still in string freeze right now for a beta release
[16:35] <cjwatson> so, difficult just now
[16:36] <davmor2> np's I was just chasing up on my posted bugs.
[16:36] <CIA-24> ubiquity: evand * r2523 ubiquity/debian/ (changelog control): Merge with doko's 1.7.13 release. Bump existing changes to 1.7.14
[16:39] <CIA-24> ubiquity: evand * r2524 ubiquity/ (configure configure.ac): Bump to 1.7.14
[16:55] <CIA-24> ubiquity: evand * r2525 ubiquity/debian/control: Update udebs, reverting doko's dropping of python-xml (that will need to be done in the individual packages)
[17:12] <cjwatson> ugh
[17:13] <evand> ugh?
[17:14] <evand> ah, that :)
[17:14] <cjwatson> yeah
[17:14] <CIA-24> ubiquity: evand * r2526 ubiquity/debian/changelog: releasing version 1.7.14
[17:17] <CIA-24> ubiquity: evand * r2527 ubiquity/ (configure configure.ac): Bump to 1.7.15
=== ceeka1 is now known as seekay
[17:48] <soren> What determines if a udeb should go an iso?
[17:49] <soren> In particular, I'm interested why the open-iscsi-udeb is not on the server iso.
[17:49] <cjwatson> the installer seed
[17:49] <cjwatson> or dependencies
[17:49] <soren> Ah.
[17:49] <cjwatson> you can actually get away with putting it in some other seed if you like
[17:49] <soren> Heh..
[17:49] <cjwatson> just please not one that maps to a task, as I'm not sure what'll happen if you do that
[17:49] <soren> Yeah, I guess I could just add it to server-ship.
[17:49] <cjwatson> right
[17:50] <soren> Cool. Thanks.
[18:03] <CIA-24> oem-config: cjwatson * r419 oem-config/ (configure configure.ac): bump to 1.30
[18:54] <CIA-24> anna: cjwatson * r407 ubuntu/debian/changelog: releasing version 1.28ubuntu2
[18:55] <CIA-24> cdrom-detect: cjwatson * r421 ubuntu/debian/changelog: releasing version 1.25ubuntu2
[21:10] <CIA-24> oem-config: cjwatson * r421 oem-config/ (debian/changelog oem-config-prepare):
[21:10] <CIA-24> oem-config: * Remove /etc/udev/rules.d/70-persistent-cd.rules and
[21:10] <CIA-24> oem-config:  /etc/udev/rules.d/70-persistent-net.rules in oem-config-prepare, in case
[21:10] <CIA-24> oem-config:  the system is going to be cloned.
[21:11] <mebrown> cjwatson, evand  just got a chance to start looking at noninteractive fixes. I am using my preseed from gutsy to test hardy and start making the changes to use the new facilities. I have found that my current preseed works fine in non-interactive, but doesnt pass partman stage (cannot find root) if I run in non-interactive mode.
[21:12] <evand> hrmm
[21:12] <evand> mebrown, can you run the install with --debug and put the logs up somewhere?
[21:13] <mebrown> where do I put --debug?
[21:14] <mebrown> Also: I noticed noninteractive mode was a tiny bit wierd and threw me off for a bit.
[21:14] <mebrown> It actually starts X and goes to the desktop
[21:14] <mebrown> and the installer continues in text mode on VT1
[21:14] <evand> ...it does?  I'll have to look into that.
[21:14] <mebrown> that really threw me off as I didnt think it was installing
[21:14] <mebrown> until I rebooted it and noticed some messages on vt1.
[21:14] <CIA-24> oem-config: cjwatson * r422 oem-config/ (debian/changelog oem-config-prepare):
[21:14] <CIA-24> oem-config: * Remove /home/oem/.ssh/known_hosts in oem-config-prepare. If ssh was used
[21:14] <CIA-24> oem-config:  then this might be slightly sensitive.
[21:15] <mebrown> This was using last week's DVD image, btw.
[21:16] <evand> so if you're booting into the livecd and then running the installer, you pass --debug to ubiquity when you call it.  If you're using ubiquity-automatic, you can use debug-ubiquity on the kernel command line as well
[21:16] <mebrown> we are using automatic-ubiquity
[21:16] <mebrown> so just add debug-ubiquity as well, right?
[21:16] <evand> indeed
[21:16] <mebrown> ok. Will try in a few mins. Need to let the current install finish.
[21:17] <evand> and then you'll have a /var/log/installer/debug file
[21:17] <evand> ok
[21:17] <evand> so /var/log/syslog, /var/log/partman, and /var/log/installer/debug will be helpful
[21:17] <evand> you'll want to strip the password information out of the latter
[21:17] <mebrown> ok.
[21:17] <mebrown> password info isnt sensitive in this case.
[21:17] <evand> ok, then dont worry about it
[21:18] <mebrown> it is 'password' and we delete that user anyways in oem-config-prepare
[21:18] <evand> ah,  indeed
[21:31] <mebrown> evand, but first....    oem-config appears broken
[21:31] <evand> oh?
[21:31] <mebrown> doh. trying to pastebin the error
[21:31] <evand> ok
[21:31] <mebrown> got cgi error
[21:31] <evand> using pastebin.ubuntu.com?
[21:32] <cjwatson> is that the zoommap error? the fix for that was only just uploaded
[21:32] <mebrown> http://pastebin.com/mfc7e25e
[21:32] <mebrown> cjwatson, yes
[21:32] <cjwatson> yeah, that's fixed in 1.30
[21:32] <mebrown> ok. I'll hold off on that, then.
[21:32] <mebrown> makes it *really* inconvenient.
[21:32] <mebrown> might I suggest an oem-config improvement?
[21:32] <cjwatson> you could upgrade oem-config* before trying to use it :-)
[21:32] <CIA-24> oem-config: cjwatson * r423 oem-config/ (configure configure.ac): bump to 1.31
[21:32] <mebrown> right now, if it fails, it opens a bash
[21:33] <mebrown> but /bin/login stomps on the same terminal
[21:33] <cjwatson> oh, damn you upstart
[21:33] <mebrown> you have to repeatedly do "open -s /bin/sh"  until bash gets the chars
[21:33] <cjwatson> yeah, it didn't when oem-config was written :)
[21:33] <mebrown> it has been this way in gutsy and hardy
[21:33] <cjwatson> since edgy
[21:33] <mebrown> ok
[21:34] <mebrown> just change the bash call to "open -s /bin/sh" or some such
[21:34] <cjwatson> I wonder what the best fix is - that's a bit tricky
[21:34] <mebrown> to have it open on an unused vt
[21:34] <cjwatson> yeah, perhaps
[21:34] <mebrown> it is just for debugging when it really craps out
[21:34] <mebrown> ok. now on to debug the non-interactive problem.
[21:35] <cjwatson> the whole panic mechanism is pretty bad really
[21:35] <cjwatson> it can't just be open -s bash since you need the error message beforehand
[21:35] <cjwatson> but yeah, something like that ...
[21:35] <cjwatson> the alternative would be to disable the login on tty1 while oem-config-firstboot is running
[21:36] <mebrown> echo MSG > /dev/tty7
[21:36] <mebrown> open -s /bin/sh
[21:36] <mebrown> or: open /bin/echo MSG
[21:36] <cjwatson> disabling the login seems safer
[21:37] <cjwatson> I don't want to assume which tty openvt is going to get, or that two successive openvt calls will get the same one
[21:37] <mebrown> or disable the login. might be easier to do with upstart
[21:38] <evand> ugh, launchpad really needs "developer" tags for comments.  That or inversely "this person doesn't know what they're talking about" :)
[21:39] <mebrown> evand, yeah, with noninteractive, it starts X for ubiquity, immediately exits X, then starts X for the desktop
[21:39] <evand> cute, I'll have a look at that later tonight
[21:41] <cjwatson> dear Scott, some job file format description would be nice, thanks
[21:42] <mebrown> evand, with noninteractive, it loops printing "No root file system: No root file system is defined"  \n, "Please correct this from the partitioning menu"
[21:42] <cjwatson> ah, moving to .bak will inhibit a job without having to delete the file entirely
[21:43] <cjwatson> mebrown: suggests a broken preseed recipe to me
[21:43] <cjwatson> s/preseed/partman preseed/
[21:43] <mebrown> cjwatson, works fine in non  noninteractive mode
[21:44] <mebrown> cjwatson, I'm willing to concede that it might be broken, I just thought it odd that it works one way but not the other.
[21:44] <mebrown> cjwatson, I'm using my base gutsy preseed as a baseline
[21:44] <cjwatson> I second the request for logs :)
[21:44] <mebrown> cjwatson, coming up. I forgot the debug-ubiquity last time
[21:44] <mebrown> rebooting
[21:49] <mebrown> -rw-r--r-- 1 michael_e_brown michael_e_brown 2934609 2008-02-29 09:47 partman
[21:49] <mebrown> it just loops...
[21:49] <mebrown> I dont see any more debug info in there with debug-ubiquity, but I can't switch VTs to VT1 now for some reason (blank screen)
[21:49] <mebrown> pastbin-ing the first part of it now...
[21:51] <mebrown> hmm... a bit big to pastebin
[21:54] <mebrown> http://fedora.pastebin.com/m2ce330e0
[21:55] <mebrown> cjwatson, evand ---^   I hope I got enough of the logfile in there.
[21:58] <evand> do you have the debug log?
[21:58] <mebrown> /var/log/debug?
[21:58] <evand>  /var/log/installer/debug, that is
[21:58] <evand> indeed
[21:58] <mebrown> no
[21:58] <mebrown> I added debug-ubiquity to the kernel cmd line
[21:59] <mebrown> do I have to *add* debug-ubiquity, or *replace* automatic-ubiquity?
[21:59] <mebrown> I added
[21:59] <mebrown> just confirmed in /proc/cmdline
[22:00] <evand> you have to add it
[22:00] <mebrown> ok. Any good reason why /var/log/installer/debug isnt there?
[22:00] <evand> and automatic-ubiquity (or only-ubiquity, but that doesn't affect you) need to be present
[22:00] <mebrown> right
[22:01] <mebrown> I have 'dm' and 'version'  in /var/log/installer/ and that is it
[22:01] <evand> hrm, if you had both debug-ubiquity and automatic-ubiquity, I cannot see why it wouldn't output the debug log.  Looks like another bug :/
[22:03] <evand> so one way around this for the time being, is to launch the regular live CD environment and call ubiquity from a terminal with ubiquity --debug --automatic
[22:03] <mebrown> ok.
[22:03] <evand> sorry, I hate to make you keep re-running the installer
[22:04] <mebrown> it isnt a problem. It is normally pretty quick.
[22:04] <mebrown> except when I forget to make my install partition active
[22:04] <mebrown> and have to boot to my usb key to fix it.
[22:04] <mebrown> normally *very* easy
[22:04]  * mebrown forgets to make install partition active and boots rescue usb key...
[22:05] <evand> heh
[22:06] <mebrown> evand, anything else on the cmdline to make it noninteractive?
[22:07] <evand> what do you mean?
[22:07] <mebrown> ubiquity --debug --automatic      ... do I need a --noninteractive as well?
[22:07] <mebrown> to get text mode
[22:07] <evand> oh
[22:08] <mebrown> its ready to go... waiting...
[22:08] <evand> ubiquity noninteractive --debug --automatic
[22:09] <evand> will give you the text based install
[22:09] <evand> otherwise you're going to get a GUI with a progress bar
[22:09] <mebrown> did I need to save its stdout?
[22:09] <evand> you shouldn't
[22:09] <mebrown> I see a ton of text
[22:09] <evand> curious
[22:10] <mebrown> and no /var/log/installer/debug
[22:10] <mebrown> but what looks like a ton of debug text on stdout
[22:10] <mebrown> and a traceback that may or may not be the culprit:
[22:10] <mebrown> debconf.py, line 81 in command
[22:10] <mebrown>   status = int(status)
[22:10] <mebrown> ValueError, invalid literal for int() with base 10: ''
[22:11] <mebrown> oh, and HAL has a field day with the repartitioning and formatting, trying to mount things.
[22:13] <evand> ah, my mistake.  Yes, the output will go to the console, though I'm not sure why I did that
[22:13] <mebrown> running it again now
[22:13] <mebrown> saving output
[22:14] <evand> you're probably better off shutting down the display manager and running it from the console
[22:14] <mebrown> yeah, about that...
[22:14] <evand> otherwise as you said things are going to get mounted a lot
[22:14] <mebrown> 'ubuntu' user will login in ... seconds
[22:14] <evand> sudo /etc/init.d/gdm stop
[22:14] <mebrown> any way to stop that?
[22:15] <mebrown> ok
[22:15] <mebrown> trying again... just a sec.
[22:15] <evand> ok
[22:17] <evand> back momentarily
[22:18] <mebrown> ok. behaviour now is different. I'm not getting the looping warnings about root fs. It just dies with the ValueError traceback I gave earlier. Looks like empty string being passed to int()
[22:22] <evand> can you put up syslog and debug?
[22:23] <mebrown> doing that now.
[22:23] <mebrown> sort of arduous to pastebin it... :(
[22:28] <evand> back in a minute, it's high time I moved this computer to a dedicated internet connection
[22:44] <mebrown> evand,  http://pastebin.com/m6592bb60
[22:44] <evand> thanks
[22:44] <mebrown> different than what I get in a direct boot
[22:46] <mebrown> just did the 'forget to activate reinstall partition thing again, and I dont have my key...'
[22:46] <mebrown> I'm going to do this again from a fresh boot
[22:46] <mebrown> as soon as I get my key
[22:56] <evand> mebrown, I think this gives me enough to work with, though this bug is separate from your rootfs bug.
[23:07] <mebrown> evand, interesting...
[23:07] <mebrown> you can take out 'automatic-ubiquity' when you put 'noninteractive' on kernel cmdline
[23:07] <mebrown> and the X crazyness I saw earlier goes away.... text only...
[23:08] <mebrown> evand, and I think that the ValueError may be what is causing the rootfs problem.
[23:08] <mebrown> because the rootfs problem comes up *immediately* after the valueerror
[23:14] <evand> ah, this explains a lot
[23:14] <evand> though he's left
[23:14] <evand> I'll investigate and then email