[00:01] ah didn't do make clean for translations and missed the error... [00:03] remove Hungarian from src/wubi/translator.nsh (line 25) and compile again [00:05] fantastic, thanks [00:06] 1.3M? [00:06] 1.1 [00:06] but r447 was 1.4M [00:06] do not know why but am glad about that :) [00:07] I had 2.1 and 1.3... [00:07] that though means that the savings are 300K as opposed to 700K [00:08] indeed, but that still helps, I imagine [00:08] thanks [00:08] ok, I'm heading out for a little bit. I'll be back later tonight. [00:09] sure [00:14] evand see my msg about umenu text for option #2 being too long [00:16] umenu rev 17 is up [00:17] did not fix the option 2 text though [00:17] 'night === TheMuso__ is now known as TheMuso [08:46] not sure if this is the spot to ask... but, I have a question regarding the installer. [08:46] I have ubuntu booting fine via tftp boot [08:47] it grabs the preseed file fine off the we server [08:47] does every thing.. but it refuses to use my internal mirror of the repository... [08:48] is the online install repo hard coded in to the installer? [08:48] here is the relevane section in preeseed: [08:48] 14 d-i mirror/http/hostname string http://horus.farorbit.com [08:48] 15 d-i mirror/http/directory string /os/ubuntu/gutsy/ [08:48] 16 d-i mirror/suite string gutsy [08:59] HADES\stephans: "http://horus.farorbit.com" is not a hostname. It's a URI. [09:03] right, that's a likely cause [09:03] soren it matches the documentation here: http://d-i.alioth.debian.org/manual/en.i386/apbs04.html [09:03] should I omitt the http://? [09:04] no, it does not match that documentation [09:04] d-i mirror/http/hostname string http.us.debian.org [09:04] http.us.debian.org is not the same as http://us.debian.org [09:04] you should definitely omit the http:// [09:04] ok [09:05] it wouldn't do any harm to file a wishlist bug on choose-mirror to have it skip over http:// automatically, mind you [09:05] right [09:48] evand, cjwatson: just to let you know that received a report of a user with working suspend under wubi/ntfs! [09:49] I must admit I didn't test the feature with the new kernels, and cannot do so until tonight. If someone else could also confirm this would be great! [10:11] also my understanding is that hibernation doesn't work with a swap file (correct me if I am wrong), but it does not seem that the pm policies check for that [10:11] i.e. you are shown the hibernate button even if your swap is on file [11:25] 2 users confirmed they can suspend... [11:31] nice [11:38] yes I am really pleased, was just asking mjg59 if there was any change in that respect! [11:39] I did squeeze down wubi by turning off debugging symbols by the way [12:03] partman-xfs: cjwatson * r755 ubuntu/ (3 files in 3 dirs): * Make finish.d/aptinstall_xfs and valid_filesystems/xfs executable again. [12:04] partman-basicmethods: cjwatson * r713 ubuntu/ (choose_method/filesystem/do_option debian/changelog): [12:04] partman-basicmethods: * When changing filesystem, set the default mount options to the contents [12:04] partman-basicmethods: of /lib/partman/mountoptions/${fs}_defaults if it exists (LP: #160450). [12:04] partman-basicmethods: cjwatson * r714 ubuntu/debian/control: Ubuntu maintainer address [12:04] partman-basicmethods: cjwatson * r715 ubuntu/debian/changelog: releasing version 38ubuntu1 [12:04] partman-basicmethods: cjwatson * r716 ubuntu/ (choose_method/dont_use/choices debian/changelog): * Make choose_method/dont_use/choices executable again. [12:04] partman-basicmethods: cjwatson * r717 ubuntu/debian/changelog: releasing version 38ubuntu2 [12:05] * cjwatson hits that cscvs bug from the dawn of time with a spoon [12:05] partman-xfs: cjwatson * r756 ubuntu/debian/changelog: releasing version 35ubuntu2 [12:07] partman-partitioning: cjwatson * r668 ubuntu/ (debian/changelog free_space/new/do_option): * Silence warning in case $id/options already exists. [12:08] partman-partitioning: cjwatson * r669 ubuntu/debian/changelog: releasing version 54ubuntu4 [12:57] partman-target: cjwatson * r707 ubuntu/ (commit.d/clear_partitions debian/changelog): * Don't clear partitions that aren't marked as using a filesystem. [13:05] partman-target: cjwatson * r708 ubuntu/ (5 files in 4 dirs): [13:05] partman-target: * Run clear_partitions from finish.d, not commit.d; commit.d is run in all [13:05] partman-target: sorts of places, such as while configuring RAID, where we don't want to [13:05] partman-target: clear partitions yet (LP: #198106). [13:05] Hi [13:05] thanks [13:06] right, the changes I've made can be replicated as follows [13:06] boot the installer, and run through it until it asks you for a partitioning method (Guided, Manual, etc.) [13:06] press Alt-F2 and then Enter to start that shell [13:06] then type the following commands exactly: [13:08] chmod +x /lib/partman/choose_method/70dont_use/choices /lib/partman/finish.d/70aptinstall_xfs /lib/partman/valid_filesystems/25xfs [13:08] mv /lib/partman/commit.d/60clear_partitions /lib/partman/finish.d/10clear_partitions [13:09] nano /lib/partman/finish.d/10clear_partitions [13:09] scroll down to line 19, where it says '[ -f "$id/method" ] || continue' [13:09] immediately below that, add a line that says: [13:09] [ -f "$id/use_filesystem" ] || continue [13:09] save and exit [13:09] return to the installer (Alt-F1) and continue [13:09] [done] [13:10] OK, giving it a try now [13:11] (you can use tab-completion to help with the awkward filenames of course) [13:14] Gah, damn vi-less installer [13:15] It failed because the 10clear_paritions included :w ;) [13:15] Trying again properly.. [13:16] I use vi myself, but memory is more important in the installer so we pick a single one that most people can cope with [13:17] OK, that worked [13:18] Only oddness is that I created two RAID partitions - a small one for /boot, larger to use as LVM [13:18] Only prompted me to create one MD device, the other seemed to already have been created [13:20] I'll try not to debug that at the same time :-) [13:20] presumably that's not actually a problem as such, just odd? [13:20] Yeah [13:22] Nice, just noticed the installer uses relatime by default now :) [13:23] Actually, there was one other small issue [13:24] The installer complained about not being able to mount /dev/md1 (which is a pv for lvm) [13:24] Let me continue fine though [13:24] I think that's a known one [13:24] IIRC anyway [13:25] there is some pain around lvm/raid combinations, but I think mostly warnings [13:25] yeah, relatime by default was yesterday [13:25] ok, that sounds good enough, I'll upload that [13:25] thanks [13:25] NP, thanks for sorting it :) [13:26] partman-target: cjwatson * r709 ubuntu/debian/changelog: releasing version 54ubuntu2 [13:28] s/yesterday/Tuesday/ [13:38] ubiquity: cjwatson * r2542 ubiquity/d-i/update-changelog: use dch -t option, added in devscripts 2.10.9 [13:39] oem-config: cjwatson * r429 oem-config/d-i/update-changelog: use dch -t option, added in devscripts 2.10.9 [14:06] xivulon: fantastic! [14:07] indeed! when you have a couple of mins can you test it to confirm? [14:09] will do === mpt_ is now known as mpt [14:29] A user reported the following issue during kernel upgrade: [14:29] ln: creating hard link `/boot/initrd.img-2.6.24-11-generic.dpkg-bak' => `/boot/initrd.img-2.6.24-11-generic': Operation not permitted [14:30] I managed to upgrade successfully yesterday night === mpt_ is now known as mpt [14:32] perhaps ntfs-3g has issues with hard links [14:34] do not recall having such issues in the past, and as mentioned I did a kernel upgrade myself which I'd expect would involve similar steps [14:35] indeed [14:40] that code in update-initramfs changed recently [14:40] perhaps the semantics for hardlinks on NTFS aren't POSIXy enough to support what we need [14:40] TheMuso_: ^-- thoughts? [14:41] I bet you can't rename over the top of another file with link() or something :-( [14:45] I really need to always configure my VM installs to output to a serial console. I just ran into a hard lock when installing Wubi, but I cannot reproduce it now. [14:50] was that during ubiquity? same amd64 issue? [14:50] didn't hear any complaint about that from users by the way [14:52] indeed, during X startup. [14:52] I'd thought that the news on suspend had to be compensated somehow... [14:53] heh [15:05] Do not think that X startup should be related to wubi though... Would be a first... [15:07] I imagine it would be a bug with ntfs-3g or FUSE if it was Wubi-specific, but I wouldn't worry about it. If I can reproduce it, I'll take care of getting decent logs. [15:12] "Fix: file and directory removal, rename and move may have been denied with an "Operation not supported" error message." - perhaps related? That's from the ntfs-3g changelog for 1.2216 [15:17] looks plausible [15:18] what is the current version of ntfs-3g? [15:19] 1.1120 [15:20] Also, I checked with the ntfs-3g ML regarding the amd64 and VMWare bug and was told by a SuSE developer that the bug (which is in FUSE, not ntfs-3g) is not present in the FUSE kernel module from 2.6.24 (instead of using the one from fuse-2.7.2) [15:21] ah good catch [15:22] well, he did basically tell me to Google next time as it was the top hit on the ntfs-3g forums for the top of the stack trace :/ [15:23] I'll take it to -kernel. [15:23] heh [15:24] would it be possible to upgrade ntfs-3g? [15:24] it might be that the error is triggered only in some particular circumstances though [15:24] which would explain why I did not notice a thing [15:24] not sure, I imagine that's a call for slangasek to make. [15:25] can you ask him or do you want me to do it? [15:25] if there's a good reason to upgrade it, we probably should - subscribe the ubuntu-release team to the bug and ask for a feature freeze exception [15:26] I'd guess the same holds for fuse/vmware [15:27] I'll write a bug report then [15:28] but if I link that to the kernel-upgrade failure, it might be difficult to prove that 1.2216 might actually fix that [15:28] since the ntfs-3g error might be sporadic [15:29] I will make it a bit more generic maybe [15:44] I have created bug #199161 [15:44] Launchpad bug 199161 in ntfs-3g "File and directory removal, rename and move may have been denied with an "Operation not supported"" [Undecided,New] https://launchpad.net/bugs/199161 [15:45] I subscribed the ubuntu release team, let me know if I need to do anything else for the feature freeze exception [15:49] ubiquity: cjwatson * r2543 ubiquity/ (d-i/manifest debian/changelog): [15:49] ubiquity: * Automatic update of included source packages: partman-basicfilesystems [15:49] ubiquity: 56ubuntu3, partman-basicmethods 38ubuntu2, partman-ext3 49ubuntu2, [15:49] ubiquity: partman-jfs 25ubuntu1, partman-partitioning 54ubuntu4, partman-reiserfs [15:49] ubiquity: 39ubuntu1, partman-target 54ubuntu2, partman-xfs 35ubuntu2. [15:55] ubiquity: cjwatson * r2544 ubiquity/debian/po/ (79 files): debconf-updatepo [16:11] ubiquity: cjwatson * r2545 ubiquity/debian/changelog: releasing version 1.7.16 [16:13] ^-- just in case they need to respin with the partman-basicmethods fix [16:14] ah, whoops. [16:16] not your fault [16:16] I had forgotten about the antediluvian cscvs bug that caused that [16:27] cjwatson: thanks for working on relatime-support.. it should be the default now? [16:28] gah, to the bus -> [16:30] would you think that a lower swappiness would be beneficial in wubi installations (given that the swap is an ntfs-3g file)? [16:30] tjaalton: alternate only for the moment, desktop support probably not in alpha 6 but in beta [17:22] cjwatson: I've done a couple of netboot installations, and at least mount(1) doesn't show the options? [17:22] -s [17:38] tjaalton: I had to add it to busybox mount [18:16] cjwatson: I mean the installed system doesn't seem to use it, or is there something missing? [18:18] tjaalton: does it show up in /etc/fstab? [18:19] cjwatson: no, only "defaults" [18:20] can I see your syslog and the date of the netboot image you used? [18:21] you need to have downloaded it on Tuesday evening or later [18:21] the image is from yesterday morning.. damn [18:21] oh [18:23] Would it be reasonable to call dhclient in the initramfs? The removal of eth0 from the interfaces file breaks preseeding via a URL on the desktop CD. [18:23] urr. might need to ask that in a wider forum, I'm not quite sure [18:23] you could do it if and only if preseeding is in use [18:24] Or perhaps I should just add the lines back to interfaces if url= is set? [18:24] cjwatson: http://users.tkk.fi/~tjaalton/foo/syslog [18:26] tjaalton: busybox definitely too old [18:26] Mar 5 15:45:53 kernel: klogd started: BusyBox v1.1.3 (Debian 1:1.1.3-5ubuntu10) [18:26] busybox (1:1.1.3-5ubuntu12) hardy; urgency=low [18:26] * Add support for relatime and norelatime mount options (closes: #460824). [18:26] -- Colin Watson Tue, 04 Mar 2008 11:02:33 +0000 [18:27] cjwatson: yeah, I'll try again tomorrow :) [18:27] maybe downloaded it from a mirror that was behind [18:27] oh, we were having internal mirroring problems yesterday morning [18:27] that would probably be it [18:28] yep [18:29] I fetch installer images from a.u.c, the rest is mirrored locally [20:32] looks like I spoke too soon on the ntfs-3g issue... The chap came back and he is using fat32 not ntfs... [20:33] so the feature freeze exception looses one case study.... [20:34] I'd still think that 1.216 is useful [20:35] but now I have to find out what is the issue with vfat [20:39] obvious q are hard links supported in vfat? [20:49] cjwatson: hrm. Interesting, but not entirely surprising to me. I wonder whether we should check what filesystem /boot is sitting on perhaps? At this point, I have no other ideas. [21:25] evand ping [21:26] mebrown: pong [21:33] we had a conversation here that I was going to ask you something, but I've forgotten... [21:34] just a sec while I think... :) [21:35] heh, ok [21:36] evand, nope, cant remember. doh. [21:38] haha, that's ok. Just ask in here or via email when you do :) [21:44] evand ^ on ntfs-3g [21:49] xivulon: hard links are not supported on vfat, as I understand it [21:50] yeah that was my guess too [21:50] I am wondering though if hard links in kernel upgrade scripts are a new feature [21:51] I don't think I have seen anyone complain about that before [21:51] probably because hardly anyone uses fat32 anymore :) [21:52] Hard links don't work on NTF either do they? [21:52] NTFs [21:54] Oh of course not, that was what was being discussed earlier. [21:55] TheMuso: works for me [21:55] evand: HRM so its just FAT32 then. [21:56] indeed [21:56] Actually... That might explain why the wubi test I did the other day on a FAT32 windows install didn't boot properly after it had finished installing. :) [21:56] xivulon: perhaps we should just tell people using fat32 to covnert their filesystem to NTFS? I can't see any reason why someone would want to keep using FAT32. [21:56] You don't support < Windows 2000 anyway, right? [21:58] evand: People still may want to access windwos from Linux without having a separate partition. [21:59] evand: Although the problem with this, is that FAT32 wouldn't support symbolic links either I would think. [21:59] TheMuso: I don't follow. Why wouldn't they be able to do that? [22:00] evand: Yeah thinking about it more, now that we have ntfs-3g with write support, it is totally nullified. [22:00] evand I did have several users with fat... [22:00] so I am a bit perplexed [22:00] symbolic links do not work for me on vfat [22:01] can you double check hard links [22:01] that is only relevant for /boot [22:01] I am not sure why hardlinks are needed at all to be honest... [22:01] I would have thought that you just have to generate an initrd and drop it in /boot [22:01] xivulon: Its more complex than that I'm affraid. [22:02] xivulon: It has to do with making sure that a valid initramfs is still present, even if a new initramfs generation fails. [22:02] https://wiki.ubuntu.com/HardyInitramfsErrorHandling [22:02] I think is the correct URL. [22:02] xivulon: double checked, no such luck [22:03] we could fall back to a more stupid and less reliable implementation if the hardlink attempt fails [22:03] cjwatson: Thats what I'm starting to wonder about. [22:03] like cp -a (takes more space, but would do) [22:05] cjwatson: I guess we could simply put the hard linking into an if check, and if it fails, then we copy... [22:06] well, ln -f blah || cp -a blah [22:06] but yeah [22:06] that would be great! [22:07] not really much that can go wrong with ln that wouldn't break cp too [22:07] aside from vfat [22:07] so that approach should be safe [22:07] * cjwatson sighs at crappy non-POSIX filesystems in this day and age [22:07] well, vfat isn't really this day and age :) [22:08] yet heaps of devices still use it. [22:08] * TheMuso looks at USB sticks. [22:08] point taken [22:08] have a look at your usb key... [22:09] Ok I will have to rectify the feature freeze request [22:09] I still think that it is valid, since ntfs shortcomings might also have produced the same issue [22:09] Well thsi initramfs issue is a bug, and well found too. [22:11] *my* USB key is ext3 :-P [22:11] reformatting that does not count... [22:12] partman-target: cjwatson * r710 ubuntu/debian/ (changelog rules): * Fix build; commit.d is gone, really. [22:13] how about I test-build partman-target this time :-/ [22:14] heh [22:15] urgh, ubiquity doesn't notice d-i components failing to build [22:15] Alright, what I might do, is make initramfs changes, and check whether they make a difference locally on a modified image using wubi. If so, I'll upload. [22:16] partman-target: cjwatson * r711 ubuntu/debian/changelog: releasing version 54ubuntu3 [22:22] f13 ping [22:22] doh. wrong channel [22:25] ubiquity: cjwatson * r2546 ubiquity/ (configure configure.ac): bump to 1.7.17 [22:58] ubiquity: cjwatson * r2547 ubiquity/ (d-i/Makefile debian/changelog): * Ensure that our build fails if one of the d-i components fails to build. [22:59] oem-config: cjwatson * r430 oem-config/ (d-i/Makefile debian/changelog): * Ensure that our build fails if one of the d-i components fails to build. [23:03] oem-config: cjwatson * r431 oem-config/ (4 files in 2 dirs): [23:03] oem-config: * Add a check-manifest tool to ensure that d-i/manifest is in sync with [23:03] oem-config: d-i/source/ on upload (from ubiquity). [23:19] What file or files does wubi check on the CD to check that it is an Ubuntu CD? [23:22] ubiquity: cjwatson * r2548 ubiquity/ (d-i/manifest debian/changelog): * Automatic update of included source packages: partman-target 54ubuntu3. [23:32] xivulon: How does wubi know what CD it is being run from, i.e ubuntu, kubuntu, etc? [23:33] TheMuso, it parses .disk/info [23:33] if it's an ISO it first extracts .disk/info [23:36] to be precise, wubi scans all disks for signs of an Ubuntu CD starting with the drive hosting the exe [23:36] if a CD is find it enters into CD "mode", so that for instance you cannot select a different distro [23:44] Ah ok, thanks. [23:50] hmm is it possible/useful to test suspend-to-ram in vm (virtualbox)? [23:50] suspend to ram seems to be disabled by default in such setup and if I run the command manually does not work too well [23:53] ubiquity: cjwatson * r2549 ubiquity/ (d-i/Makefile debian/changelog): [23:53] ubiquity: * Don't clean console-setup; bits of it are intentionally missing from our [23:53] ubiquity: copy, and it's not necessary. [23:54] oem-config: cjwatson * r432 oem-config/ (d-i/Makefile debian/changelog): [23:54] oem-config: * Don't clean console-setup; bits of it are intentionally missing from our [23:54] oem-config: copy, and it's not necessary.