[00:04] evand, well i think it looks like bug 223671. i've marked it accordingly. once gnome-python-desktop is split up ubiquity's depends can be fixed up [00:04] Launchpad bug 223671 in ubiquity "wnck and rsvg should be provided in seperate packages not requiring gnome" [Undecided,New] https://launchpad.net/bugs/223671 [02:22] cjwatson: around? [02:36] joshk: He would be asleep at this time. [02:36] I suspect [02:36] ok [02:41] i guess i'll send him an email [08:39] argh, completely forgot about that dependency [08:40] I'll just rework it to completely use PNGs, as previously planned [09:27] Did mathiaz post his ISO testing scripts somewhere? [09:54] soren: Not sure, but perhaps this is it: https://code.edge.launchpad.net/~mathiaz/+junk/iso_testing_scripts [10:07] stgraber: I'll need the syslog as well as the preseed file [10:07] evand: You must be one of those Google users. [10:07] :) [10:07] evand: Thanks. :) [10:07] anytime [10:38] I'm trying to test my solution for bug 291670, and keep running into the installer failing to detect the CD-ROM drive under kvm. The first advice I received indicated that this was probably transient, but I've replicated on a few dailies. Is this an issue with the kernel, with the installer, or something else? (read: how should I file a bug) [10:38] Launchpad bug 291670 in base-installer "LPIA installer missing kernel" [Medium,Triaged] https://launchpad.net/bugs/291670 [10:42] persia: doesn't happen to me on i386. I'm guessing that the kernel packaging is failing to include the necessary modules in the udebs it ships [10:42] No, it's specific to the lpia alternate CD. I'll chase up with the kernel folk. Thanks. [10:42] persia: use lspci to figure out the device, correlate against kernel source to figure out which module is involved, and then look in debian/d-i/ [10:43] persia: "kernel packaging failing to include necessary modules in udebs" is exactly the sort of thing that is specific to alternate CDs, and could well be lpia-specific, so ... [10:43] it *could* be an installer bug, but that sounds less likely in this case [10:53] persia: If you need to send a patch to the kernel folks, either Colin or myself would be happy to go through sending one to them which they can apply right out of the mail. [10:55] TheMuso, Thanks for the offer, but I'm not sure I'll get that far. To me, the kernel still remains the territory of grues. [10:55] persia: in the case of d-i, it is very easy. [10:55] Its just adding a module name to a file. [10:57] Ah. Yes, that I think I can do :) The phrase "patch to the kernel" always interrupts my thinking processes a bit :) [10:57] heh [10:57] debian/d-i/ is not really the kernel. It used to be in a separate set of packages but got imported into the packaging. [10:58] s/packaging/kernel packaging/ [11:01] I guess the only sticking point is dealing with git. [11:02] Well, at this point, the sticking point is tracking down what's missing :) [11:03] Once that's sorted, whether it's git or a patch in a bug, or just barking at someone is the simple bit. [11:07] * TheMuso nods. [11:23] cjwatson: Regarding https://wiki.ubuntu.com/FoundationsTeam/Specs/OemTrackingId , do you think there should be a default DCD, and if so, any suggestions for the value? ubuntu-cdimage was my thought. [11:49] evand: I don't think there should be a default DCD [11:49] ok, noted [12:02] evand: When is the new timezone map going into ubiquity? [12:05] cjwatson, evand: I seem to be getting this issue just recently http://www.davmor2.co.uk/partition.png [12:05] shouldn't the images stay on the one page? [12:07] you mean the scrollbar? [12:08] it looks like it couldn't make the text fit *shrug* [12:10] cjwatson: I mean the hd representation at the top [12:12] Is that literally because the text is too long? [12:17] looks like it, yes [12:18] maybe ought to wrap or something [12:18] cjwatson: Ah okay I'll have a look against vista at some point and see if it is the same [12:29] partman-auto: cjwatson * r281 ubuntu/ (71 files in 25 dirs): merge from Debian 84 (the hard way, since it was released from the lenny branch) [12:34] partman-auto: cjwatson * r282 ubuntu/debian/changelog: releasing version 84ubuntu1 [12:40] partman-auto-lvm: cjwatson * r207 ubuntu/ (4 files in 3 dirs): merge from Debian 32 (the hard way, since it was released from the lenny branch) [12:42] partman-auto-lvm: cjwatson * r208 ubuntu/debian/changelog: releasing version 32ubuntu1 [13:07] davmor2: that's a known bug [13:07] that will be fixed Real Soon Now (tm) :) [13:11] evand: Real Soon Now (tm) isn't that short before the next LTS :) [13:13] * cjwatson kicks partman-auto-lvm. Stupid overcomplicated code [13:20] heh [13:56] Hello all. Anyone online? [13:57] better to ask your question [13:57] okay. [13:58] I've just set up a laptop with intrepid via a network installation, but for some reason, the keyboard isn't in English. It appears to be Arabic or some such. [13:58] I.e. I can't type in anything in English characters. [13:58] I'd like to see /etc/default/console-setup please. Also does this affect console or X or both? [13:58] console, I can't even get into X, yet. [13:59] try alt+shift (and release) and see if that switches to English [13:59] Is there a way to change the keyboard config on the fly? [13:59] nope [13:59] if you can access the system remotely you can change /etc/default/console-setup and run 'sudo setupcon'. But I'd like a copy of the file first [13:59] also was this an automatic installation of some kind? [14:00] Yes. I used a local laptop as the tftp/pxe boot/preseed server [14:00] Also running Ubuntu 8,.10 [14:00] then perhaps the preseed file would be useful to me [14:01] What's the best way to get it to you? [14:01] https://bugs.launchpad.net/ubuntu/+source/console-setup/+filebug [14:01] make sure to erase passwords from it, obviously [14:17] okay. i just submitted bug 328078. The preseed.cfg file is attached. [14:17] Launchpad bug 328078 in ubuntu "Ubuntu network installer sets incorrect keyboard locale" [Undecided,New] https://launchpad.net/bugs/328078 [14:24] you didn't use the URL I gave though ;-) [14:24] (reassigned to console-setup) [14:25] redmage123: so the main thing I notice is that you haven't actually preseeded keyboard configuration [14:25] redmage123: you've used preseeding information from some Debian documentation, or perhaps some very old (< edgy) Ubuntu documentation [14:26] isn't d-i console-keymaps-at/keymap select us going to give me the proper keyboard setup? [14:26] no; that's based on a keyboard component in the installer that was removed in edgy [14:26] oops. [14:26] https://help.ubuntu.com/8.10/installation-guide/i386/appendix-preseed.html is the documentation to use for 8.10 [14:26] https://help.ubuntu.com/8.10/installation-guide/i386/preseed-contents.html#preseed-l10n advises "d-i console-setup/layoutcode string us" [14:27] it would be nice if we had a more graceful default, though, and I've said as much in the bug [14:29] the reason we changed the preseed variable name rather than maintaining compatibility, by the way, was that we switched to generating console keymaps from X keymaps, and so you now select an X keymap name rather than a console keymap name [14:29] in some cases these overlapped (e.g. us), but in many cases they didn't, and I think there were even some conflicts [14:29] so it was better to have a clean break [14:30] got it. I'm trying the new config now... [14:31] I do have to say that setting up an automated network installation of Ubuntu has been...challenging. :-) [14:31] For some reason, even though I had mounted the intrepid iso on my local system, I simply coudln't get the remote installer to use it. [14:31] it's probably easier if you follow the right documentation ;-) [14:32] so, I had to switch it to an ubuntu installation site on the internet. [14:32] we don't generally support installing from a mounted ISO; while ISO images contain *most* of what you need for a netboot installation, they really aren't guaranteed to do so [14:32] the netboot initrd is starting out from a different set of installer components [14:33] that said, it ought to be possible, I just tend not to have very much sympathy when it breaks ;-) [14:34] Okay, so what if I'm in a situation where, for whatever reason, I need to install ubuntu over the network but don't have internet access or have a heavy firewall presence which I can't modify? Am I SOL? [14:35] for firewalls, I generally advise (getting somebody who can to) create a local mirror. You're going to need one anyway for installing any other packages after the fact or for security updates [14:36] Is there any good docs for creating a local mirror? [14:36] still, it may be possible to get it to work in your case; I'd be happy to look at a log of it failing to use the mounted ISO image to figure out why it didn't work [14:37] debmirror(1) [14:39] I don't think it's necessary to debug the iso issue, this is really more of my lack of experience in doing network installs in Ubuntu than anything else. I just assumed (wrongly) that pointing remote install to the .iso was the correct procedure. [14:44] it's probably just incorrect preseeding somewhere ... [14:44] "d-i mirror/country string manual" rather than "enter information manually" might help [14:44] although shouldn't be relevant [14:45] d-i partman-auto/init_automatically_partition \ [14:45] d-i partman-auto/disk string /dev/sda [14:45] that's bizarre - that expands to "d-i partman-auto/init_automatically_partition d-i partman-auto/disk string /dev/sda" which I'm sure isn't what you meant [14:46] might be less effort in the long run to start from scratch with the Ubuntu example preseed file :-) [14:52] cjwatson: we're opening a bug report now with everything attached [15:00] cjwatson: bug 328097 [15:00] Launchpad bug 328097 in debian-installer "preseeding partitionning isn't working anymore hardy 8.04.2" [Undecided,New] https://launchpad.net/bugs/328097 [15:12] evand, well splitting python-gnome2-desktop is feasible still, does rsvg provide enough advantages (scaling) that it would be more worthwhile to use? [15:14] It's somewhat hard to say as the requirements have changed slightly as the code has been developed, so while rendering SVGs as the mouse moved was basically unusable, we now only render the highlight on click. Still, I'm inclined to go with PNGs just because I've been bitten so badly by the poor performance of the old design. [15:14] It's not going to be much work to make it use all PNGs (as I'm partly through it now). [15:14] oh yeah definitely agree there then [15:14] good deal [15:14] stgraber: any chance you could do a bit of set -x debugging work for me? [15:15] stgraber: remove enough of your preseeding so that you get the hostname prompt, then run through the installer until the hostname prompt appears, switch to tty2, 'nano /bin/perform_recipe', and put 'set -x' on the second line [15:16] stgraber: then switch back to tty1, continue the installation, and send me the syslog again [15:16] (I've looked through the existing logs and they don't really give me enough, unfortunately) [15:22] cjwatson: sure [15:24] cjwatson: I don't have perform_recipe when at the hostname prompt [15:24] isn't it downloaded right after that prompt ? (I'm using mini.iso) [15:26] can you stick around for a bit? I have to run right now [15:27] sure, I'm EST so I have plenty of time :) [15:46] cjwatson: re NSLU2: Issue was that the ramdisk was overwriting APEX when unpacked; we found values which work but need a change in APEX; ogra experiments a bit more to find optimal values for APEX and we hope to get these changes in Debian [15:58] lool: ok, cool [16:40] ubiquity: evand * r3021 ubiquity/ (27 files in 3 dirs): [16:40] ubiquity: Make the timezone map only use PNGs to avoid a dependency on [16:40] ubiquity: python-gnome2-desktop and because we want to optimze for speed rather [16:40] ubiquity: than image quality. [16:40] ^ Apologies for growing the size of the bzr branch by removing and adding all of those images. [17:17] evand, can you upload that change today too so that xubuntu and mythbuntu dailies aren't broke? [17:38] I'm having real problems installing ubuntu on my aspire one netbook. I've tried making usb key installers from iso images using unetbootin, but they result in boot errors on the aspire. the .img file for UNR succeeds until it gets to "will mount root from /dev/sdb". Any ideas? [17:38] is it just that the usb keydrive is rubbish? I've tried 2 [17:38] are there any ubuntu net install images (.img files) that i could use, as I have more success with those? [17:42] have you tried usb-creator? that's the one we ship by default [17:43] well on my desktop machine i run plain debian and its not available in the apt-get sources which i use. [17:44] tunk: netbook is an .img file [17:44] davmor2, and it does boot, but it gets stuck when it gets to "will mount root from /dev/sdb" [17:44] so for me, while it looks amazing, UNR is not working [17:44] I suspect the UNR guys would love to know about that; most of them are not here [17:45] we don't ship USB installation images of pure Ubuntu, only the application, since unfortunately it would basically double space requirements on our CD mirrors and that's a serious problem [17:46] is it possible for me to create an .img file and then write it using dd? I think for UNR it's also as good to take Intrepid and install the packages on top. [17:47] i guess that's what usb-creator does anyway. [17:48] so do the UNR guys hang out in here? [17:48] tunk: #ubutnu-mobile [17:48] thanks [17:56] superm1: absolutely [18:00] ubiquity: evand * r3022 ubiquity/ (d-i/manifest debian/changelog): [18:00] ubiquity: Automatic update of included source packages: partman-auto [18:00] ubiquity: 84ubuntu1. [18:32] ubiquity: evand * r3023 ubiquity/debian/changelog: releasing version 1.11.8 [18:44] cjwatson: hi, you there? [18:46] joshk: yeah, for about five minutes [18:47] d'oh. will you be back? [18:48] silly UTC [18:49] well, lemme make it quick [18:49] <-- GET passwd/user-password-again [18:49] --> 1 [18:49] less says.. *^C [18:49] I'll be back in several hours [18:49] ok [18:49] blink. *memory corruption*? [18:49] it's all sh and python ... [18:50] well, apart from cdebconf. but that hardly ever has problems [18:50] oh, but this is ubiquity so it's debconf, ergo perl [18:50] hmm, wait, never mind. turning on the debug stuff messed with my generated preseed file [18:50] my bad [18:50] ah [18:50] come back later, i'll probably have something for you then [18:50] mkay [18:51] you're slightly reliant on my evening time, but I'll see what I can do :-) Failing all else I do read scrollback [18:51] okay [18:51] thanks [19:01] stgraber: I think the best approach, then, might be to use expert mode (priority=low), to step through the installer until just before the partitioner starts, make those changes, and then set the priority back up to high or critical or whatever you want from the relevant main menu item [19:20] cjwatson: so, it was showing me the wizard because of user-setup/encrypt-home being unset [19:20] should there be a default for that? [19:20] or should it be a hard option [20:29] Hello all. [20:29] So, I'm still running into this bizarre problem where the keyboard thinks it's in Arabic. [20:30] Can I change this from the xdm menu? [20:30] I tried setting the language to US/English, but it doesn't seem to make any difference... [21:52] joshk: hmm. that alone perhaps shouldn't trigger it [21:52] redmage123: just out of interest, are you preseeding console-setup/layoutcode as a kernel parameter, or only in your preseed file? [21:53] cjwatson: if i preseeded it, the install goes on [21:54] right. I think I would be inclined to call that a bug, on balance [21:54] together with auto-login [21:54] though I confess I'm not immediately sure how to fix it given the general architecture [21:55] need to divide up the question set for each page into ones that must be shown or preseeded and ones that can use the default - sort of a priority distinction I suppose [21:57] redmage123: the thing is that the console is configured *before* the preseed file is processed (along with the locale and other such early bits of the installer), so you generally need to set that on the kernel command line. Sorry, I should have mentioned this earlier. console-setup/layoutcode=us is the way you do this as a boot parameter [21:57] well, a band-aid fix for now would be to default it to 'no' [21:57] or something [21:58] the code in question is shared with the alternate installer, so we actually don't have a particularly good place to handle defaulting for that right now [21:58] I'd like a bug about this, and will try to figure it out [21:59] (ubiquity runs user-setup behind the scenes and intercepts the debconf protocol; any time it gets an INPUT that isn't preseeded, that causes it to stop on whatever page it's on and ask the user) [22:00] hmm, actually maybe it isn't too difficult [22:01] cjwatson: Do I set this up in the pxelinux.cfg/default file? [22:01] cjwatson: against ubiquity, then? [22:01] joshk: yes please [22:01] redmage123: yes [22:01] k, will do [22:01] redmage123: in the append line [22:02] joshk: I think I have a handle on how to fix it but a bug will be useful retrospectively ("why the hell did I do that?") [22:02] joshk: /ubuntu/+source/ubiquity, that is, not /ubiquity [22:02] It's already in there. The append line looks like: [22:04] append ramdisk_size=14984 locale=en_US console-setup/layout=en_US... [22:04] that must be console-setup/layoutcode=us, not console-setup/layout=en_US [22:04] aha. [22:04] en_US isn't a valid keyboard layout name, so it guesses wildly [22:04] same for the locale directive? [22:04] no, en_US is a valid locale [22:05] just that locales and keyboard layouts are different things :) [22:05] us is lower case or is the parameter case insensitive? [22:05] lower case [22:05] I don't remember offhand, but lower case will definitely work [22:05] Thanks. I'll retry this... [22:05] was there some documentation that led you to do console-setup/layout=en_US? If so, I'd love to get it fixed, because you aren't the first person who's made this mistake [22:06] and I have a suspicion that this is on some web page somewhere ... [22:06] I'm going to definitely do a installation writeup and post it somewhere for others. [22:06] (it does need to be layoutcode rather than layout BTW - arcane internals) [22:07] Yup. Got it. [22:07] BTW, one thing I'd love to see in an installer is the ability to do an install and break into a shell rather than do the full install. Is that possible? [22:08] for things like recovering a system? [22:08] or something else? [22:08] I hate it when I do a complete installation and then realize I've messed something up and have to reinstall. [22:08] i'd love to be able to get a certain way into an installation and then be able to break into a root shell rather than have to go through the whole installation again. [22:08] Does that make any sense? [22:08] I'm sort of wondering what you would do in the root shell [22:09] just stop and have a look around to check that things are sane? [22:09] it would be a bit time-consuming, but you could run with priority=medium and then you could step through the install piece by piece [22:09] Possibly, or the ability to fix a specific thing that i've screwed up in the installation. [22:10] and there are shells on tty2 and (in recent versions) tty3 that you can use [22:10] if you know the installer well enough it's entirely possible to modify it on the fly from those shells [22:10] (installer developers do it all the time ...) [22:10] Well, I do have to say that i've learned a tremendous amount about ubuntu installation in the last couple of weeks. :-) [22:11] the database of questions and answers that everything lives in is accessible from those shells (debconf-get, debconf-set), although there are some slightly awkward restrictions about setting things from there - workaroundable if you know exactly what you're doing [22:12] what you're asking for sounds like something I do pretty regularly, i.e. run through the installer a little way to check that it's done the right thing. Going back is sometimes tricky for one reason or another but is usually possible *somehow* [22:14] joshk: you know, I'm really stupid - this is a one-liner. I just need to *remove* those two questions from the list we pay attention to [22:15] okay, well, you can just look busier then by closing the bug :) [22:15] joshk++ [22:17] ubiquity: cjwatson * r3024 ubiquity/ (debian/changelog ubiquity/components/usersetup.py): [22:17] ubiquity: Don't show the "Who are you?" page just because the auto-login or [22:17] ubiquity: encrypt-home questions are asked and not preseeded (LP: #328281). === kirkland` is now known as kirkland [22:32] \o/ [23:18] redmage123: any luck confirming that proposed fix? [23:29] the installation is still underway. it's kinda slow. Currently it's in "Select and install software" [23:29] six per cent done. [23:29] Been like that for a little while. [23:34] * cjwatson blames apt [23:34] I think that's it sitting preconfiguring packages or something, but without a progress bar [23:43] It's gonna take probably another half hour or so before it's finished, I think.