| maxb | Hello, can anyone suggest further reading after https://wiki.ubuntu.com/UbiquityAutomation for more details? | 07:24 |
|---|---|---|
| mattcen | GrueMaster: Thanks for your input earlier, but unfortunately my preseed already has the two lines you suggested. Back to square one. | 07:29 |
| mattcen | maxb: Assuming you read the debian-installer preseed guide, I'd imagine that'd be all you need? I don't use ubiquity so can't be of any further help. | 07:30 |
| maxb | The wiki page is a little thin on the details. I guess I can always go read the ubiquity source code | 07:31 |
| xnox | maxb: well, we embed d-i, so some d-i things work. But most plugins, self-evidently in the source code read values from debconf. | 08:12 |
| xnox | maxb, what are you after? | 08:12 |
| maxb | Well, apt sources.list customizations first | 08:13 |
| maxb | I could just go straight to the source code, I just wanted to double check there wasn't any more detailed documentation first | 08:15 |
| xnox | maxb: there is this: http://www.debian.org/releases/stable/i386/apb.html.en but note that some things are not done via d-i, but re-implemented/adjusted to use squashfs installation. | 08:17 |
| xnox | maxb: these should work https://help.ubuntu.com/13.04/installation-guide/i386/apbs04.html#preseed-apt | 08:18 |
| maxb | Yeah, I've worked with d-i preseeding before, it's the differences from that that I'm most interested in | 08:18 |
| xnox | maxb: well see configure_apt in scripts/plugininstall.py & there are also target_hooks that can be run. | 08:24 |
| maxb | ok, thanks for the pointer | 08:25 |
| Peace- | guys i have disabled secure boot uefi and i have activated legacy boot | 09:18 |
| Peace- | i was doing the partitions | 09:18 |
| Peace- | and i got | 09:18 |
| Peace- | http://wstaw.org/m/2013/05/23/plasma-desktopaL6776.png | 09:18 |
| Peace- | so now what should i do? without ruin windows stuff | 09:19 |
| cjwatson | Create that separate partition | 09:20 |
| Peace- | ok i will try | 09:20 |
| cjwatson | The dialog is obscuring enough of your partition table that I can't see whether you have space for it without resizing/deleting something else | 09:20 |
| Peace- | ok new screenshot | 09:21 |
| peace-vaio | cjwatson: http://wstaw.org/m/2013/05/23/plasma-desktopOG6776.png | 09:22 |
| cjwatson | Yeah, you'll have to resize something to make room | 09:23 |
| peace-vaio | ok | 09:23 |
| cjwatson | As the dialog says, it only needs to be 1MB | 09:23 |
| cjwatson | Well, you might be able to squeeze it in between sda10 and sda7 | 09:23 |
| Peace- | cjwatson: but i need to use ext4 ? point of mount ? | 09:24 |
| cjwatson | I forget exactly how the KDE frontend displays it, but hopefully you'll see a "Use as" dropdown with "Reserved BIOS boot area" as one of the options, as per the dialog | 09:24 |
| cjwatson | No, follow the instructions on screen :) | 09:24 |
| cjwatson | Don't make things up | 09:24 |
| cjwatson | "Reserved BIOS boot area" | 09:24 |
| cjwatson | If it doesn't say that (which would be a bug), look for "bios_grub" | 09:25 |
| Peace- | ok ok i found | 09:25 |
| Peace- | :D | 09:25 |
| Peace- | cjwatson: thank you very much | 09:28 |
| cjwatson | np | 09:28 |
| Peace- | cjwatson: btw if i have uefi disabled windows doesnt' boot | 09:33 |
| Peace- | i hope that grub will fix that for me :D | 09:33 |
| cjwatson | you can't just rip uefi out from under Windows 8 and expect it to work | 09:33 |
| cjwatson | grub can't fix that for you | 09:34 |
| Peace- | xD | 09:34 |
| cjwatson | (never mind that grub in raring has some problems chain-loading Windows 8 in uefi mode, but that's a different matter ...) | 09:34 |
| Peace- | so i need to enable it to choose windows ? | 09:34 |
| cjwatson | yes | 09:34 |
| Peace- | oh ok | 09:34 |
| cjwatson | assuming it's Windows 8 anyway | 09:35 |
| cjwatson | but it probably is on a UEFI machine | 09:35 |
| Peace- | i asked just to know cuz windows 8 is a crap system | 09:35 |
| Peace- | i don't think i will boot to windows so often | 09:35 |
| cjwatson | you seem to have made some trouble for yourself by disabling uefi entirely. why not just disable the secure boot part, assuming that was the bit that really bothered you? | 09:35 |
| cjwatson | (though it's probably more trouble now to go back and redo it ...) | 09:36 |
| Peace- | cjwatson: i was trying to boot insto kubuntu | 09:37 |
| cjwatson | note that there are systems that have trouble booting OSes from GPT in BIOS mode - hopefully yours isn't one, but they do exist; the ones I've seen can still boot but just won't do it automatically | 09:37 |
| cjwatson | ah, granted, we still need to fix Kubuntu for UEFI. saucy might be happier there, needs testing | 09:37 |
| Peace- | so it didn't work with uefi , disabled via bios => switched to usb legacy somethin like that and it worked | 09:38 |
| cjwatson | this came up at UDS. it's a little mysterious why it failed; as far as I can see the only thing missing for Kubuntu was the signed kernel | 09:39 |
| Peace- | cjwatson: anyway i did 4 restoring dvd for windows ... | 09:39 |
| cjwatson | but that isn't supposed to be a failure | 09:39 |
| maxb | I've been attempting to use qemu to test an ubiquity installation, but it seems like qemu's SDL display doesn't want to display the graphical environment. Is this a known issue? | 11:43 |
| xnox | maxb: try a different graphics? -vga should work correctly. | 11:44 |
| maxb | "-vga std" / | 11:45 |
| maxb | "-vga std" ? | 11:45 |
| maxb | Done of the documented -vga options seem to work. | 11:46 |
| maxb | It works through the syslinux part of the boot, but as soon as it tries to boot linux, I just end up with an unresponsive console with a cursor flashing. If I wait a bit, the window resizes, but it still displays just a text mode cursor | 11:48 |
| maxb | I suppose VNC might work, though that requires running a separate client - more difficult to launch from a script | 11:48 |
| === kentb-out is now known as kentb | ||
| GrueMaster | mattcen: Morning. I'll look and see what I can come up with. One thing you can do is do a manual install, then get the d-i settings used from a d-i dump, although it is quite ugly. Lot of extra "stuff". | 15:03 |
| GrueMaster | Although someone with inside knowledge (aka cjwatson) might know more and have better answers. | 15:05 |
| cjwatson | Pbas1: Hm, this really isn't very clear ... | 15:50 |
| cjwatson | Looks like netcfg is exiting non-zero for some reason, but I'll have to insert a bit more instrumentation to see why | 15:51 |
| Pbas1 | K thxs a lot (I spend 2 week to try to resolve that ... without success). | 15:55 |
| Pbas1 | but the install cd must working because I generate a correct one but ... I can't remember how | 15:57 |
| Pbas1 | as soon I generate Packages.gz and Release it doesn't work anymore. | 15:57 |
| cjwatson | Yeah, I'm still looking | 15:57 |
| cjwatson | Certainly doesn't seem to be doing much at the load-cdrom stage, which is suspicious | 15:59 |
| cjwatson | Pbas1: You seem to be missing a bunch of the stuff I'd expect to be here; for example you have no bootstrap-base package listed | 16:04 |
| cjwatson | Oh | 16:04 |
| Pbas1 | debootstrap? | 16:04 |
| cjwatson | Yes, you have completely fouled this up I'm afraid :-) | 16:05 |
| cjwatson | /dists/lucid/main/binary-i386/Packages* lists a load of installer modules (*.udeb) | 16:05 |
| cjwatson | All the *.udeb entries must be in /dists/lucid/main/debian-installer/binary-i386/Packages* | 16:05 |
| cjwatson | So you need to look at why whatever you did to regenerate the Packages files got that wrong | 16:06 |
| cjwatson | The way it is, basically half the installer's brain is missing | 16:06 |
| Pbas1 | My iso contain /cdrom/pool/main/d/debootstrap/debootstrap-udeb_1.0.20ubuntu1.4_all.udeb like the ubuntu server CD | 16:11 |
| Pbas1 | and I see udeb file in Package* | 16:12 |
| ogra_ | in /dists/lucid/main/debian-installer/binary-i386/ ? or in dists/lucid/main/binary-i386/ ? | 16:12 |
| ogra_ | :) | 16:12 |
| ogra_ | you want the former | 16:13 |
| ogra_ | (or d-i does) | 16:13 |
| Pbas1 | both | 16:13 |
| Pbas1 | I agree :-) | 16:13 |
| cjwatson | Pbas1: The contents of pool/ are not what I'm talking about | 16:14 |
| cjwatson | Pbas1: You have some udebs referenced in /dists/lucid/main/debian-installer/binary-i386/Packages, but not all. I don't know why. | 16:14 |
| cjwatson | Pbas1: A lot of important ones are in dists/lucid/main/binary-i386/Packages instead. | 16:14 |
| cjwatson | Pbas1: And that's your problem. | 16:15 |
| Pbas1 | Ok then I can try to correct with apt-ftparchive from the top level of pool | 16:16 |
| cjwatson | Happy to look at your apt-ftparchive configuration. | 16:16 |
| cjwatson | If you run it with the same configuration a second time it'll presumably still just get it wrong in the same way. | 16:16 |
| Pbas1 | yes but my packages* will list all my udeb (my kernel 3 udeb and the old one) | 16:18 |
| Pbas1 | I don't use override option with apt-ftparchive I just lanch this command: | 16:18 |
| Pbas1 | apt-ftparchive packages ./pool/main/ |gzip > dists/stable/main/binary-i386/Packages.gz | 16:19 |
| Pbas1 | apt-ftparchive packages ./pool/main/debian-installer/ |gzip > dists/stable/main/debian-installer/binary-i386/Packages.gz | 16:19 |
| cjwatson | Yeah, that's completely wrong. | 16:19 |
| Pbas1 | no surprise | 16:19 |
| cjwatson | There's no need to put your udebs in their own directory in the pool, and you've confused yourself badly by doing so. | 16:20 |
| Pbas1 | I separate the installer kernel3'udeb files that I generate. | 16:20 |
| cjwatson | I understand what you've done, but my point stands. | 16:21 |
| Pbas1 | I will copy all my udeb in pool/main then | 16:21 |
| cjwatson | I don't think it's possible to do this correctly with apt-ftparchive's "packages" subcommand. | 16:21 |
| cjwatson | Or, well, you could but you'd have to filter the resulting files afterwards. | 16:22 |
| cjwatson | Better to set up a configuration file and use "apt-ftparchive generate". | 16:22 |
| cjwatson | Then you can use Packages::Extensions entries in the appropriate sections to select .debs or .udebs. | 16:22 |
| Pbas1 | I tried this option but I failed with override option (that I don't understand) | 16:23 |
| cjwatson | It could use examples, really. Let me try to write you something more correct | 16:25 |
| Pbas1 | Thxs I tried without success with this doc: https://help.ubuntu.com/community/InstallCDCustomization#Building_the_repository_with_apt-ftparchive | 16:26 |
| cjwatson | Yeah, that was what I was about to recommend, actually. Please tell me *specifically* (with a transcript) what went wrong when you tried that | 16:28 |
| Pbas1 | the result was wrong :-( : the same thing =< the installer stop after the network config. | 16:29 |
| Pbas1 | the override file didn't contain 3.0 kernel ref anyway | 16:30 |
| cjwatson | Shouldn't need to | 16:30 |
| cjwatson | Can you please prepare a set of configuration files according to that doc, and send me a tarball of them? What you're trying to do here is get me to the point where I can reproduce your problem. | 16:31 |
| cjwatson | Because I'll probably be able to find a small tweak that *does* work. | 16:31 |
| cjwatson | But it would take me quite a while to write it all from scratch, and it sounds like you've tried most of it already ... | 16:31 |
| Pbas1 | yes but I deleted that: my forge is a virtual machine... | 16:32 |
| Pbas1 | but I can begin that again. | 16:32 |
| cjwatson | Thanks. The configuration files in that doc look roughly right to me, and they'd be where I'd begin to save time in any case | 16:33 |
| cjwatson | There are various subtle issues that can arise from leaving out the override files for things in our archive (although as I say they shouldn't make much difference for the kernel .udebs you're adding), so I wouldn't like to write something that omits them. | 16:34 |
| Pbas1 | all my new udeb are in pool/main/debian-installer/ | 16:36 |
| cjwatson | Irrelevant | 16:36 |
| cjwatson | The structure of the pool is not supposed to matter | 16:36 |
| cjwatson | Not at this level, anyway | 16:37 |
| cjwatson | It only matters because of the particular apt-ftparchive commands you're using - which are problematic in any event, so it's better to come up with something correct | 16:37 |
| cjwatson | Like I say, I'm happy to help given a starting point, just don't have time to prepare the entire thing from scratch | 16:37 |
| Pbas1 | I didn't know if there are predefined structure | 16:38 |
| Pbas1 | K I will prepare a nice tarball | 16:38 |
| cjwatson | The dists tree has structure that matters, but in general pool is only supposed to be inspected by way of the Filename references in Packages/Sources files | 16:38 |
| maxb | Is it possible to disable the initial "press a key" boot screen with the (choices) = (keyboard) icon, and go straight to the syslinux menu? (perhaps by changing syslinux.cfg?) | 17:13 |
| cjwatson | That's governed by the hidden-timeout setting in gfxboot.cfg | 17:15 |
| maxb | thanks :-) | 17:15 |
| maxb | Hrm, I'm trying to enable oem-config by preseeding oem-config/enable to true, and it *is* causing the installer to mention OEM mode in the window title, but it's not actually leaving the new system set up for booting into the temporary OEM user | 17:20 |
| cjwatson | Today's image? | 17:21 |
| maxb | cjwatson: raring release | 21:31 |
| maxb | So my problem with oem-config seems to have been that I confused it by preseeding a user name | 22:33 |
| maxb | Except without preseeding a user name, I seem to be unable to convince it to skip displaying that page of the installer prompts | 22:34 |
| === kentb is now known as kentb-out | ||
| maxb | Hrm. It seems if I preseed them to the same values oem-config-udeb sets anyway, it skips the page | 22:44 |
| maxb | Argh. It looks like a side effect of bug 1167622 is that raring installation media is not reliably usable for EFI installations | 23:07 |
| ubot2 | Launchpad bug 1167622 in linux (Ubuntu) "Cannot change EFI variables using efibootmgr (raring regression)" [Medium,Confirmed] https://launchpad.net/bugs/1167622 | 23:07 |
| maxb | Is there any Ubiquity way to install extra packages? An analogue to pkgsel/include ? | 23:35 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
