[00:09] I'm doing a netboot install, and it hangs at 75% of "Storing language..." but I can still use another console. Is there anything useful I can do? (Oh, and #ubuntu is nearly useless.) (and cjwatson: I got 6.06LTS installed on my 486 box with 36MB RAM. now trying on a P/166 with 80MB RAM) [00:09] er cjwatson_ ^ [00:10] I can switch to a console. Anything useful I can do there? [00:12] last pertinent log message was "localechooser: en_US.UTF-8..." about ten hours ago. [01:04] ok maybe this one: when trying to netboot hardy heron, it says there are no kernels to install. tips or pointers to info. much appreciated. [02:41] hi: when i install gutsy, reboot and try to login using gdm i get you dont have home dir create so you have to use / (root) like home dir, why i have to create home dir manually? [07:35] hi all [07:36] I'm not given a resize option in the paritioning part of a gutsy install [07:36] is this normal? [07:51] ah! [07:51] sudo gparted :-) [08:29] simeon: not necessary [08:29] simeon: I assume you're talking about the autopartitioner; in that case the automatic resize option can only be offered in some quite limited circumstances due to how crap the PC partition table format is [08:29] simeon: but if you select manual partitioning you can change sizes there; there's no need to resort to gparted [08:29] ah [08:30] cjwatson: well I also wanted to resize the NTFS partition, so gparted saved me lots of ntfsresize man page reading [08:31] the manual partitioner would have let you resize NTFS too [08:31] you created lots of work for yourself, it sounds ;) [08:31] oh :-) [08:31] okay [08:31] seems I didn't have enough faith in the installer [08:31] wasn't too tought though [08:32] gparted, click, click, resize, done, run installer [08:33] yeah, I really dislike people having to rely on gparted though [08:33] we put a lot of work into replacing it for a reason :) [08:36] to be honest, I did a bit of googling and I saw some (possibly older) screenshots showing an option to resize the NTFS parition on the first paritioning screen [08:37] you know, the one with, o - guided use whole parition, o - manual paritioning [08:37] and mine didn't have that [08:37] right, that's what I was saying above [08:37] it can only be offered in certain quite limited circumstances [08:37] we plan to expand those a little bit in hardy, but they'll still be limited [08:37] and NTFS+SATA is one combination where it doesn't work? [08:37] no, it depends on the partition table layout [08:38] as in the number of primary partitions you have, the position of the extended partition, that sort of thing [08:38] SATA has nothing to do with it [08:38] right [08:38] if the NTFS partition is unresizable for some reason, then that would also kill it, but since you resized it with gparted I doubt that's the case [08:38] in gutsy, if you have three primary partitions already, it won't offer auto-resize [08:39] (three primary partitions and no extended, that is; or four primary partitions) [08:39] ah, HP recover and HP tools paritions :-) [08:39] yeah, I'm beginning to hate vendors ;-) [08:40] like I say, I think we can relax this a bit in hardy due to fixes made elsewhere [08:40] partman-auto-loop: cjwatson * r36 ubuntu/ (autopartition-loop debian/changelog): * Account for partman library code having moved to /lib/partman/lib/. [08:45] partman-auto-loop: cjwatson * r37 ubuntu/debian/changelog: releasing version 0ubuntu12 [08:48] cjwatson: I assumed that since I did not get the resize option on the initial screen that a manual paritioning would just change the table, and not resize the FS [08:49] cjwatson: but maybe a "typical" user does not know the difference between resizing a parition and the underlying FS [08:49] HCI is fun :-) [08:51] manual partitioning would change both; changing the table without changing the filesystem is an absurd division of functionality to present in a UI :) [08:58] I won't argue with that [08:59] but I should mention that many other installers will do exactly that [08:59] especially older ones [09:00] so I suspect many users have come to expect the installer to do something absurd :) [09:27] hardy still fails to netboot here.. I'll post the log shortly [09:31] tjaalton: same thing as before? [09:31] http://users.tkk.fi/~tjaalton/foo/syslog [09:31] no, different [09:31] but I've seen this as well [09:31] 403 [09:31] current apt? [09:31] ah [09:32] try now [09:32] the mirror is uptodate, so is the netboot image [09:34] you seem to have ended up with 'stty olcuc' somewhere in the middle of that [09:34] apt is current [09:34] heh, yes :) [09:34] Jan 10 09:20:50 in-target: E: [09:34] Jan 10 09:20:50 in-target: Command line option 'i' [from -icons] is not known. [09:34] that doesn't ring any bells [09:34] I'll check the preseed [09:37] hm, there are no changes though, and it's mostly the same preseed with feisty which works [09:37] oh right [09:37] can I see the preseed file? [09:37] found it, typo in pkgsel/include :/ [09:38] that'd do it :) [09:38] dodgy line breaking? [09:38] cjwatson, evand, would it be possible to take care of the wubi tagged bugs? All the ones that require coding should have a patch attached and/or link to a branch [09:39] it would semplify my life quite a bit and it is a necessary to eliminate the lupin-support package [09:40] cjwatson: we use the same package list for 7.10 and hardy, which obviously fails by now, so I've added rules to filter out some packages, and it fails since some of the packages had a "$pkg-icons" variant on the list :) [09:40] xivulon: not today but yes [09:40] cjwatson thanks, feel free to ask if you need clarifications or require some more work on those issues [09:40] it would be nice if aptitude would be more graceful when it cannot find all the packages on the list [09:41] xivulon: you should default to evand not me [09:41] s/would/could/ [09:41] * xivulon pokes evand [09:41] tjaalton: it's worse than that in this case because it's being interpreted as an option [09:41] yep, true [09:41] + * Pass pkgsel/include to aptitude following "--" to prevent typos in [09:41] (committed upstream) [09:41] + preseed files being interpreted as aptitude options by accident. [09:42] heh [09:43] right, fvwm is uninstallable, so fvmw-icons becomes "-icons", duh [09:43] err [09:43] I doubt that [09:44] I think it's more likely you have a space in there by mistake [09:44] oh, unless you were doing s/fvwm//g? [09:44] yep [09:44] ah, ok [09:44] so s/fvwm //g works for now :) [10:14] cjwatson forgot to check that, but does the live cd initrd prompt the user to insert a CD if one live media is not detected? Don't think so. [10:16] no, I think it loops waiting for one and eventually panics [10:17] if you plan to keep using wubi-cdboot it would be a good idea to add that prompt [10:17] I can file a bug report === cjwatson_ is now known as cjwatson [10:24] xivulon: I don't know; that has a nasty interaction with systems where it takes ages for the CD device to appear normally [10:39] cjwatson: it seems that all the logs entries of packages being installed are in caps [10:46] yes, I'm mystified but not concerned [10:47] tjaalton: Is this amd64? [10:48] soren: yes [10:48] tjaalton: I'm seeing the exact same thing in my amd64 sbuild (not i386)... I've spent a bit of time tracking it down, but haven't figured it out yet. [10:48] It's been that way since gutsy. [10:49] soren: cool, as said it's not that important, but annoying :) [10:49] It doesn't happen on my real system, or if I schroot into an amd64 chroot and apt-get install stuff. Only if sbuild does it. [10:49] hmm, my gutsy install log seems fine [10:50] My best guess was that something accidentally sent stty olcuc to my terminal, but AFAICS that wouldn't case the logs to be in uppercase, too. [10:52] unless it's going through a virtual tty [10:52] True. [11:46] hmm, casper still preseeds the mouse protocol? [11:46] on livecd [11:46] (EE) Configured Mouse: Unknown protocol "10 xserver-xorg/config/inputdevice/mouse/protocol doesn't exist" [11:47] $ wcgrep mouse/protocol [11:47] $ [11:47] that looks to me like xorg trying to fetch a question that doesn't exist [11:48] right, but it doesn't use that anymore [11:48] dexconf or the postinst [11:49] hopefully casper.log will show the culprit [11:49] bug 181732 btw [11:49] Launchpad bug 181732 in xorg "[hardy][livecd] xorg configuration problems on qemu-kvm" [Undecided,Incomplete] https://launchpad.net/bugs/181732 [11:49] hardy netboot finished fine once I filtered all the failing apps from pkgsel list [11:51] $ wcgrep mouse/protocol [11:51] ./debian/local/Failsafe/failsafeDexconf:169:fetch xserver-$SERVER/config/inputdevice/mouse/protocol [11:51] ./debian/xserver-xorg.postinst.in:920: "xserver-xorg/config/inputdevice/mouse/protocol" \ [11:51] oh right :/ [11:51] umm [11:51] current version in hardy [11:51] the second one is cleanup code [11:51] yep [11:51] the first is your bug [11:52] presumably the failsafe copy didn't get fixed ... [11:52] right, but the real bug is why it got there :) [11:58] to failsafe-mode that is [12:50] cjwatson can you elaborate on "nasty interaction"? Worse case scenario is that you prompt the user to insert a CD even if it is already inserted at which point the user will probably hit enter [12:51] hmm, I suppose [12:51] Which I think it's better than showing them the busybox prompt [12:52] bug #181734 [12:52] Launchpad bug 181734 in casper "Prompt the user to insert a CD if a live media is not detected" [Undecided,New] https://launchpad.net/bugs/181734 [13:39] xivulon: sure thing, I'll take a look at them today [14:36] right, though today is alpha3 so I cannot release any of those changes. [15:23] evand, I can grab you on jabber and we can go through together [15:25] if possible I'd like to have them by the w/e so that I can test my stuff [15:26] PS I am having a "chat" (on forum) with henrik re WinFOSS/launcher in CD [15:26] The idea is to replace it with a small nsis app that displays 4 options [15:27] http://people.ubuntu.com/%7Ehenrik/images/wubi-mockup.png [15:27] xivulon: Sure, but can we defer that to say tomorrow after today's alpha release? Henrik emailed me about the forum thread. I believe I have a call with him later today. [15:27] http://ubuntuforums.org/showpost.php?p=4108900&postcount=8 [15:27] thanks [15:28] evand, sure I'll try to grab you tomorrow night (feel free to start merging stuff before then) [15:29] I'm definitely going to look over them today to prepare for the discussion tomorrow. [15:30] https://bugs.launchpad.net/wubi/+bugs?field.tag=wubi [15:30] indeed, I already have it opened in a tab :) [15:30] I have another grub-install patch to submit, will do that tonight [15:31] ok, I'll keep an eye out ofr it [15:31] for* [15:31] good news is that half of them are taken care of by ~ago/partman-auto-loop/lupin-support [15:47] by the way is it possible to mount hardy squashfs in gutsy? [15:51] why would you want to do that? [15:58] mmh that's probably the right channel :) [15:58] hello mates [15:58] I was wondering if there's a way to use a response file with ubuntu graphical installer [15:58] could you confirm it or give me some pointers ? [15:59] what is a "response file"? [15:59] do you mean preseeding the answers to certain questions? [15:59] yes [15:59] as of 7.10, yes; generally speaking the question names are the same as for d-i, with a few exceptions [16:00] any documentation on this ? [16:00] evand: ? [16:00] I can't find some anywhere [16:00] ^-- expert [16:00] okay ;) [16:01] * evand makes standard comment about needing to write that documentation. Adds it to today's TODO. [16:01] I found this but it seems to apply to D-I [16:01] https://help.ubuntu.com/community/InstallCDCustomization [16:01] one moment [16:01] not ubiquity, right ? [16:02] the ubiquity automation is designed to be compatible where possible, so a decent starting point is to copy a d-i preseed file [16:03] okay, so all I need to do is to check ubiquity help and launch it with the right argument and a DI preseed file ? [16:04] the preseed argument is fished out of /proc/cmdline since that's how people usually like to start ubiquity [16:04] so url=http://blah usually [16:04] okay :) [16:05] boot on the regular ubuntu live cd [16:05] add the right kernel flag to start ubiquity at boot time [16:05] and url parameter to the pressed file [16:07] automatic-ubiquity being that kernel parameter. [16:09] so boot the Ubuntu desktop CD and press F6 at the islinux menu, add url=http://evalicious.com/evan.seed automatic-ubiquity to the argument list, as an example. [16:09] nice :) [16:09] choises that are not presseded will be asked to the user ? [16:09] feel free to use that preseed file, but be forewarned, it will format your hard drive without warning [16:10] s/choises/choices [16:10] correct [16:10] nice [16:10] this's all I need [16:10] perfect :) [16:10] * evand makes note to correct the languagechooser question in that file. [16:10] thanks a lot for your help [16:10] let me know if you run into any trouble [16:10] yeah I'll have some time to make test and do exactly what I want [16:11] a last thing, do you confirm it's only available since 7.10 ? [16:11] correct [16:11] okay :) [16:11] thanks! [16:11] you're welcome [16:23] a last one I just thought about [16:23] ubiquity by pxe ? [16:23] possible ? [16:27] ping evand ;) [16:29] can't you use the alternate installer for that? [16:29] I could do so [16:29] but the aim of the project is to provide a very simple way to install ubuntu for noobs users [16:29] so I'd avoid the old curse dialogs.... [16:31] seems we share the same aim... [16:32] netbooted ubiquity has been done in the past though may need a bit of plumbing [16:32] cjwatson if that was available I could add that to wubi also [16:32] any pointers would be greatly appreciated [16:32] as we're now two people that need it [16:32] :p [16:33] xivulon: it only makes sense with local mirrors; I'd prefer not to offer it to users without proper IT-department-type backing [16:34] what the good way to start working on that ? [16:34] trying to install something like the livecd in a nfsroot [16:34] and boot it by pxe ? [16:35] I think passing NETBOOT=nfs to the live CD boot process is the way to start [16:35] but I haven't done this personally [16:36] err, and NFSROOT=server:/path [16:36] do you mean loading livecd kernel by pxe and passing the right nfsroot as kernel arg ? [16:36] yes [16:36] that was the kind of stuff I was thinking about [16:36] kernel and initrd [16:37] the only real problem is how to turn a regular ubuntu deboostrap [16:37] just unpack the squashfs and export that over NFS [16:37] oh, really ? [16:37] don't try to recreate the squashfs contents if you don't have to [16:37] ugh, I hate this wireless card. Sorry about that. [16:37] mount -oloop the iso, copy it and export it by nfs ? [16:39] mount -o loop iso mntpoint && mount -o loop -t squashfs mntpoint/casper/filesystem.squashfs export-mntpoint [16:39] but COPY the contents [16:40] I would not advise NFS-exporting something that's actually a squashfs mount point - it may well work but why make your life harder? [16:44] mount iso + mount squafs over, and then copy all that stuff to the exported directory ? [16:49] yeah [16:49] not saying this will work unmodified but it should be a good start :) [16:50] yeah I think that way should work [16:51] initrd is ready to nfsboot [16:51] so I guess it should do the trick [16:54] cjwatson: Maybe it's a false memory, but isn't there somewhere where the latest installation-guide package is built and placed on ubuntu.com? I poked around archive.u.c and cdimage.u.c and grepped through my IRC logs, but came up empty. [16:54] it's supposed to be on help.ubuntu.com [16:54] Indeed, but that's stuck at 7.05 [16:54] err 7.04 [16:55] oh [16:55] right, I mailed mdke about that a little while ago [16:55] ok [16:55] he acknowledged it but hasn't fixed it yet [16:56] ok [17:21] okay :) [17:21] I managed to boot into the livecd throught pxe [17:21] but evrything fucked up [17:21] the main problem, I guess, is that I should mount a virtual ramfs over the nfsroot [17:26] casper should do that ... [17:26] apt-get source casper # place to start debugging [17:33] maybe the problem is just that nfs-common and portmap are missing into livecd squash... [17:34] sounds possible [17:36] just added them and tried a restart [17:37] got stuck on starting nfs-common [17:37] weird [17:39] still the same breakage finally [17:39] i/o error on all binaries [17:40] First cut: https://wiki.ubuntu.com/UbiquityAutomation [17:40] I guess / was not mounted magically :p [17:40] I'll work on this tomorrow :) [17:40] cya people :) [17:41] goodbye [17:41] best of luck [17:43] I'm not really sure where to link to that though. w.u.c/InstallerTeam? [17:43] via w.u.c/Ubiquity maybe [17:44] indeed, but what links to that? I can't seem to find the command in Moin. [17:49] I'd do a text search and weed out the FPs I think [17:50] but add a link to Ubiquity from InstallerTeam if it's not there already? [17:50] will do [17:53] lupin/casper provides a fourth way to pass a preseed file by the way: on a local hard disk [17:54] xivulon: feel free to add it to the wiki page [17:56] he the interface is not yet stable enough I guess, I'd suggest that we merge lupin/hardy play with it, then we write the spec [17:57] ok [17:58] bug #144798 partially implemented in lupin/hardy/casper [17:58] Launchpad bug 144798 in casper "Merge lupin functionality + add external hooks" [Wishlist,Confirmed] https://launchpad.net/bugs/144798 [18:16] evand: perhaps note on that page that ubiquity asks a subset of the same questions, rather than the same questions [18:16] since we only use some of the same components [18:17] ah, indeed, thanks for spotting that [18:19] fixed [18:46] cjwatson: does this look like a reasonable solution to the inifinte loop bug? http://pastebin.ubuntu.com/3447/ [18:47] evand: yeah, I think that's basically what localechooser does [18:47] whoops, sticking a break in the right place [18:48] it is [18:48] looks fine [18:48] ok. I imagine it's far too late to get this in tribe 3, right? :/ [18:53] s/tribe/alpha/ [18:53] I think so - but ask slangasek [18:55] err whoops, heh [18:55] will do [19:01] ubiquity: evand * r2402 ubiquity/ (debian/changelog ubiquity/components/language.py): [19:01] ubiquity: * Correctly answer both countrychooser/shortlist and [19:01] ubiquity: countrychooser/country-name (LP: #174937). [19:54] https://bugs.launchpad.net/dell/+bug/181857 [19:54] Launchpad bug 181857 in oem-config "No input methods active for firstboot -- cannot enter chinese" [Undecided,New] [19:55] cjwatson, as discussed a couple days ago. I have the system sitting at firstboot now and i cannot enter chinese (gutsy) [19:56] now, it may just be user-error, but I tried ctrl-space, shift-space, and numerous combinations thereof, and was unable to get it into chinese mode [20:06] note: both shift-space and ctrl-space work ok after finishing and logging in. [21:37] for ubiquity preseeding, does ubiquity preseed/late_command work? [21:37] or is it d-i? [21:39] ah, that's only in d-i. Use ubiquity/success_command [21:39] err [21:39] does that run inside /target or outside? [21:40] outside of /target, but it's still mounted when that runs [21:40] okay [22:03] can i suppress the apt-setup/security-updates note? [22:03] er [22:03] can i suppress the apt-setup/security-updates-failed note? [22:03] or whatever it might have been changed to in ubiquity [22:05] looks like it's still the same [22:09] Is 'ubiquity apt-setup/security-updates-failed note ' not working? [22:12] oh, haven't tried it [22:34] hm, that seemed to work [22:34] k, last question: is there a way to suppress "Close the tray and press ENTER to continue?" [22:40] hmm, there doesn't appear to be currently, but I thought we handled this use case previously. [22:41] kind of odd to allow ubiquity/reboot to be preseeded and not this :) [22:42] indeed :) [22:44] I'll fix it [22:47] ah, we did [22:47] or rather cjwatson did [22:47] add noprompt to the kernel command line [22:49] hot [22:49] michael_e_brown: right, like I said the other day, there's no input method (scim) running [22:49] cjwatson, correct [22:49] scim is the thing that handles shift-space et al and if it isn't running then it simply doesn't work [22:49] cjwatson, that would be a good thing to fix for Hardy [22:50] any way to hack it in for Gutsy? If I modified the script to start an input method, would that just work? [22:50] I dont have a feel right now for how big of an issue this is now. [22:50] I wont find out until next week. [22:52] the place to try to hack it in would likely be oem-config-dm [22:52] but I simply don't know for sure because I haven't tried [22:52] and it's 22:52 here :0 [22:52] :) [22:56] cjwatson, np. I'll find out next week and work on it if the China team thinks it is a big priority. [23:01] huh [23:02] it's okay to escape # signs in debconf preseed values with \#, right? or is there a different way to do it [23:13] joshk: hmm, that's a difference between debconf and cdebconf I hadn't noticed before [23:13] joshk: in cdebconf, only # at the start of a line introduces a comment [23:14] joshk: in debconf, AFAICS there is no way to escape # at all :( [23:14] $line=~s/\#.*$//; [23:15] oh, crap [23:17] hm, i can work around it