[00:00] cjwatson, I was mostly blindly attempting to make the resulting install similar to the results of having used moblin-image-creator, rather than thinking carefully. passwordless sudo would make more sense in that context. [00:01] mdz happened to mention it today in a different context (oem) and I thought it made sense [00:03] MIC does something subtly different though. It has the crypt of '' in ume-config postinst and sets it [00:03] (Which I find hideous) [00:03] My only fear would be with the GUI sudo proxies : it's useful to have the popup alerting the user that the following action is an admin action, even if it can be dismissed with the enter key, but perhaps that's better done through integration with some policykit frontend. [00:03] StevenK: not that far off what casper down [00:03] does [00:04] StevenK, That's essentially what user-setup does with the appropriate preseeding now. [00:04] persia: it's not exactly very obvious that you have to hit enter though; we get questions about that kind of thing from time to time when it rears its head [00:05] cjwatson, True. [00:05] * persia adds writing a spec about a good way to handle that to the TODO list for next week [00:11] cjwatson: casper is a special case, it's a live system. No fair for an installed one. :-) [09:38] TheMuso, hi, did you by any chance try wubi with sw raids? I still have in the guides "forget it", can I now change that? [09:58] oem-config: cjwatson * r555 oem-config/ (debian/changelog scripts/tzsetup): [09:58] oem-config: Use OVERRIDE_SHOW_ALL_LANGUAGES in the tzsetup wrapper as well as [09:58] oem-config: localechooser (LP: #290580). [10:13] superm1: ^- oem-config r555: we'll get this into -updates at this point, I think - is that OK for you? [10:13] sorry this wasn't caught 'til late [10:19] xivulon: Sorry I didn't try it, and can't spare a box to be cleared and set up with Windows at this point either unfortunately. [10:21] xivulon: I can try when I get back to the states. [10:23] this is strange.. netbooting intrepid for the first time, and doing "d-i preseed/include_command echo foo/bar" results in html code in the preseed log, and the install fails. this works on hardy though [10:35] oem-config: cjwatson * r556 oem-config/ (configure configure.ac): bump to 1.54 [10:46] ubiquity: evand * r2930 ubiquity/ (configure configure.ac): Bump to 1.10.9 [10:46] yep, it's the -c option for wget that breaks it [10:48] ubiquity: evand * r2931 ubiquity/ (debian/changelog ubiquity/zoommap.py): [10:48] ubiquity: Check to make sure a gdk window is mapped before calling [10:48] ubiquity: cairo_create (LP: #208800). [10:58] sigh, why do I get 416 with the -c option [11:01] usb-creator: evand * r54 usb-creator/ (4 files in 3 dirs): [11:01] usb-creator: * Internally represent the persistent file size in bytes for greater [11:01] usb-creator: accuracy. [11:01] usb-creator: * Move self.pipe declaration to the correct location. [11:01] usb-creator: * Add a bootloader installation progress message. [11:02] usb-creator: * Install the bootloader to the MBR as well (LP: #273477). [11:11] duh, so no ifconfig in busybox anymore, that's what broke my preseeding [11:21] TheMuso, evand, not in a rush, it's only to update the website/guides accordingly, users will ask me whether raids are supported or not and atm I have no clue. [11:21] ideas how to get the MAC-address inside busybox?-) [11:21] without ifconfig [11:22] evand are you in london? will be around for the release party? [11:23] tjaalton: cat /proc/net/arp ? [11:23] xivulon: ja and ja [11:23] will you be there? [11:23] hopefully yes, (if works allows) [11:24] good deal [11:25] evand: tried that, but doesn't have anything about the host [11:26] oh well, I could just tell it on the cmdline and parse that [11:33] tjaalton: use ip [11:33] busybox hasn't had ifconfig for ages [11:33] at least not in our config [11:33] ip addr show or some such [11:42] ip a l === bananin is now known as Guest97884 === Guest97884 is now known as Basti_dash [11:58] cjwatson: yes, will try that. hardy did have ifconfig though :) [11:59] mm, Debian turned it off yes === Guest97884 is now known as Basti_dash === Guest97884 is now known as Basti_dash [13:39] persia: do you have any more idea about what bug 288320's about? [13:39] Launchpad bug 288320 in ubiquity "Networkless Ubuntu MID install hangs at "Scanning archive..."" [Undecided,Confirmed] https://launchpad.net/bugs/288320 [13:40] cjwatson, It's about trying to install a language pack when it's not available and the network isn't present. [13:41] I can generate some logs, but when I was looking at it on Friday, I was tempted to blame anna. [13:42] lool wanted me to document some workarounds, but none of them were ever applied, so it's outstanding. [13:42] The proper solution is to integrate with debian-cd, rendering it much more unlikely to reach that state. [13:42] oh, this is due to lack of .disk/whatever? [13:42] maybe not [13:43] the installer really shouldn't hang in this case even if language packs aren't on the disk [13:43] can I reproduce this in an emulator without an actual device? [13:43] From what I saw, it looked like it was looping on trying to install the language pack and failing. [13:43] that's really odd [13:43] won't be anna [13:43] If you turn off networking, you should be able to do that. Alternately, install it on an old i686 without networking. [13:43] cjwatson: Yes [13:44] anna installs udebs and language packs are debs [13:44] (you can reproduce in emulator) [13:44] persia: sorry, never done this before, don't suppose you can point me to full instructions? [13:44] cjwatson, Which route? burning the image for non-network hardware, or disabling networking for kvm? [13:44] cjwatson: kvm -m 512 -cdrom ubuntu-mid.img works [13:45] lool, even with network? [13:45] err -hda [13:45] * persia is confused, and tries that, wondering if there are different error messages [13:45] lool: ok, thanks [13:45] -net none I guess [13:45] That ought do it. [13:45] Yeah, exactly [13:45] lool, You reproduced without -net none? [13:46] I was looking up the option [13:46] persia: No; I know one can use the image in kvm, and I'm confident that you can reproduce the no net issue in kvm as well [13:46] I never tried both, but I'm confident it works :) [13:46] lool, OK. That makes me feel better. [13:47] downloading [13:48] Actually, you probably want `kvm -hda ubuntu-mid.img -hdb target.img -m 512M -net none` (as I hadn't tested -cdrom before, and it's not working for me now) [13:48] cjwatson: There's a catch, you can't boot from hdb with the image, and if you boot from hda with the image and then install to hdb, it will corrupt your hda img [13:48] But you should be able to at least go as far as reproducing the issue I guess [13:48] yeah, not bothered about booting the whole thing [13:48] why will it corrupt hda? [13:49] I thought evand's changes to not break USB sticks fixed that. Double-checking (I've been playing with hardware recently) [13:49] It used to corrupt hda because it would install grub on hda when installing the target to hdb [13:50] persia: Oh it's fixed? [13:50] And Install is under Preferences, once you boot. [13:50] lool, I thought so : double-checking now. [13:50] cjwatson: If you meet the corruption issue, you can use -cdrom isolinux.iso with http://people.ubuntu.com/~lool/isolinux.iso [13:50] and select "2" to boot from hdb [13:51] persia: Saw some untranslated strings go through [13:51] "Calculating size to skip copying" or somethign like that [13:52] in ubiquity [13:52] Before "Copie des fichiers" "copying files" [13:52] lool, There's a few of those, which depends on the language. [13:52] Ok [13:52] persia: These are marked for translations but weren't? [13:52] * persia defers to someone who knows which strings are in ubiquity (rosetta-translated) and which are from d-i (d-i translated) [13:54] "Miroir de l'archive Ubuntu corrompu" (ubuntu archive mirror corrupt) [13:54] Popup with "Ok" as the only choice, and displayed in a loop with -net none [13:54] So I confirm I can reproduce in kvm :) [13:54] This is at the "Analysing mirror" step (Analyse du miroir) [13:56] lool: "Calculating size to skip copying" -> we had to put that in untranslated post-RC [13:56] lool, "Bad archive mirror" isn't language-specific. [13:56] I don't care about untranslated strings at this point anyway [13:56] it's too late to do anything about them [13:59] I just wanted to mention it in case it was a specific string which was missing _() [13:59] But it seems it's not and is known :) [14:00] persia: Isn't language-specific? Sorry don't understand what you mean [14:00] persia: Oh it's the C version of my string, ok [14:01] lool, Although I had a successful English install previously on hardware, I just ran an English install in KVM -net none and got the error, which surprises me, as my hardware has never had a net connection that worked. [14:01] persia: Black magic I tell you [14:03] persia: So, corruption or no corruption? [14:04] Hrm? [14:05] persia: I think you were checking whether hda was still corrupted [14:06] lool, Yes, but I was caught by 288320, which I hadn't seen in English before. [14:08] INPUT critical mirror/bad seems to be the interesting bit. [14:16] persia: I'll have to defer looking at this, something else urgent has come up [14:17] cjwatson, Not a problem. [14:26] persia: No, still corrupted [14:26] lool, Hrm. Oh well. I thought it worked once. [14:35] lool, Found it. I'll update the bug when I'm out of the meeting, but the issue is that apt-setup thinks we're doing a netboot install when we install from USB (as it's not marked as a CD install), and so expects choose-mirror to get a mirror from the network. [14:36] lool, Considering that this infrastructure won't be used for Jaunty, I'd say it's worth a release note at best. [14:37] ah right [14:37] so indeed, .disk on the image [14:38] persia: k === Guest97884 is now known as Basti_dash [15:15] cjwatson, its fairly less than ideal.. so if there is still an opportunity for re-rolling at least DVD disks with just this change, that'd be much more preferable [15:15] surely the absence of an "updated" live DVD throughout the last month due to archive inconsistencies played into not catching this kind of bug [15:18] superm1: my worry is that that would require respinning alternate images too, or else break jigdos [15:18] superm1: if we get this in ASAP after release, we could do a special DVD build for you ... [15:19] if I respin alternate images as well as desktop/DVD, the QA team are going to be even more heavily loaded up and I'm worried they won't make it [15:19] would they necessarily need to be retesting "everything" with the new DVDs if you know the only delta was an oem-config deb that sat in pool/ though? [15:19] it's not, we also have a ubiquity change [15:19] oh i see.. [15:20] and as I said if I uploaded oem-config then they'd have to retest alternate too [15:20] we do need to do a smoke-test of everything [15:20] at some point we will write up a list of the things that have gone wrong in the past, in order to give people an idea of the kinds of things that can happen if you try to skimp on this [15:21] well we've got a framework to allow dropping in updated debs as necessary in case things like this happen. so long as it's in -updates we can pull it into it through that framework. [15:21] I wouldn't ask the QA team to skip this, and I wouldn't expect them to say yes if we tried [15:22] just using it adds more factory burn time, so that's hwy it's preferable to have the one on the DVD go with it [15:23] I'd be happy to do a custom DVD for you after 8.10 [15:23] with just this change [15:23] okay, that'd be sufficient [15:23] ubiquity: cjwatson * r2931 intrepid/ (debian/changelog scripts/install.py): [15:23] ubiquity: Always copy the kernel from /cdrom/casper/ if it's there regardless of [15:23] ubiquity: whether any kernels are already present in /target/boot/, since those [15:23] ubiquity: could just be due to a preserve-home install (LP: #287747). [15:24] ok, great === Guest97884 is now known as Basti_dash [15:29] ubiquity: cjwatson * r2932 intrepid/debian/changelog: releasing version 1.10.9 [15:35] ubiquity: cjwatson * r2932 ubiquity/ (debian/changelog scripts/install.py): merge from intrepid branch [16:59] is grep -E supported in the installer? I want to do: grep -E (name|MHz) /proc/cpuinfo >> $TARGET/etc/motd [17:00] I am assuming /proc/cpuinfo exists [17:06] CarlFK: yes [17:06] quote '(name|MHz)' thus though [17:06] /proc/cpuinfo exists [17:07] quotes - right. thanks [18:09] old P3 box that may have hardware issues, but...: ibex alt-installer: update-apt-xapi[26136]: segfault ... kernel: in python2.5[8048000+fc000 [18:12] actully that is the 2nd segfault - 35 lines of syslog: http://dpaste.com/87601/ [18:13] not an installer bug [18:13] I am guessing this means "bad hardware" Oct 29 17:35:34 kernel: [ 1787.160872] res 51/40:00:a7:2a:47/00:00:00:00:00/e2 Emask 0x9 (media error) [18:13] looks like I/O errors on whatever ata1 is [18:13] sda I suspect [18:17] yeah - lots of Oct 29 17:07:00 kernel: [ 72.894627] res 51/84:00:28:00:00/00:00:00:00:00/e0 Emask 0x10 (ATA bus erro [18:17] grep error syslog |wc [18:17] 61 [18:17] never mind :) [19:00] ubiquity: cjwatson * r2933 ubiquity/ (12 files in 5 dirs): merge from intrepid branch [19:18] ubiquity: cjwatson * r2934 ubiquity/ (configure configure.ac): bump to 1.10.11 [20:33] ubiquity: evand * r2935 ubiquity/ (debian/changelog ubiquity/debconffilter.py): [20:33] ubiquity: Reset next_go_backup after encountering a question marked as seen [20:33] ubiquity: (LP: #290760). [20:46] user-setup: evand * r126 user-setup.ubuntu/ (debian/changelog user-setup-apply): Do not force chpasswd to md5 crypted passwords (LP: #51551). [20:50] migration-assistant: evand * r93 migration-assistant/ (debian/changelog ma-script-utils): Do not force chpasswd to md5 crypt passwords (LP: #51551).