[01:52] casper: cjwatson * r631 trunk/ (debian/changelog scripts/casper): [01:52] casper: Tail casper.log and show its messages during boot (thanks, Michal [01:52] casper: Suchanek; LP: #363886). [03:14] casper: cjwatson * r632 trunk/debian/changelog: releasing version 1.177 [03:45] If I do a three-disk install with 3-node RAID1 /boot and the rest RAID5 and LVM, grub is used. [03:45] Is grub installed on all three disks, or only the first one? [03:45] (This is with d-i from the ubuntu-server 8.04.1 CD.) [07:35] <_ruben> just the first disk, from what i recall [07:36] Is that considered a bug? [07:36] <_ruben> could be === evand1 is now known as evande === evande is now known as evand1 [10:48] cjwatson: Have you ever used the web frontend for Debconf? [10:50] for about five seconds once several years ago [11:01] cjwatson: Ok. I have a web(-like) interface from whence I'll need to let a user configure packages (by answering debconf questions). I was thinking of embedding the debconf web frontend (with some CSS applied to it, so it's less hideous) in an iframe. Another alternative would be to somehow arrange for debconf to speak directly to the my interface and basically implement a new frontend in some sort of AJAX-y fashion. [11:02] Which makes you cringe less? :) [11:02] the former [11:04] Good :) [11:10] ..although a debconf "frontend" that simply tunnels debconf somewhere else is somewhat interesting. You could be managing a server over SSH, and get asked debconf questions on a local GNOME debconf frontend. [11:10] cjwatson: Do you know if anyone has tried something like that? [11:11] "tunnels debconf" as in "tunnels the debconf wire protocol", I mean. [11:21] it's called "passthrough" [11:21] ish, anyway ... [11:21] I don't know that anyone's tried the specific thing you suggest although I believe all the pieces are there [11:23] Oh, yes, there it is. Interesting. [11:32] passthrough is used very heavily by the installer for various nefarious purposes [11:32] which all require keeping extremely careful track of which file descriptors go where, often on paper :-) [11:42] hi [11:42] i'using preseed to install ubuntu on a pool of machines [11:43] I would like to set up a few repositories inside preseed [11:43] jerroome: as in, extra lines in /etc/apt/sources.list [11:43] ? [11:44] I use apt-setup/local0/repository for the first [11:44] I saw you could use apt-setup/local1 etc [11:44] yep [11:44] but the second isn't taken into account [11:45] at least, that's what it looks like [11:45] err, really? could I see your preseed file (feel free to obscure passwords) and the syslog from the installation attempt? [11:45] (if the installation completed, it'll be in /var/log/installer/syslog) [11:46] but it doesn't complete, it fails while installing packages [11:46] should I paste my preseed file here ? [11:46] please don't paste things into IRC [11:46] use paste.ubuntu.com or similar [11:46] you can paste something there and it'll give you a URL, which you can paste here [11:47] ok, just give me a sec [11:52] here it is http://paste.ubuntu.com/188147/ [11:55] jerroome: what version of Ubuntu are you installing? [11:55] jerroome: it looks OK, but can I see your installer syslog so that I can look for errors there? [11:56] as the install fails, I can't give it to you [11:56] that simply isn't true [11:56] you can use 'save debug logs' from the installer main menu to extract it [11:56] it is based on 8.04 server [11:56] or you can run 'anna-install openssh-client-udeb' from a terminal, and then you can scp the file to another machine [11:57] in the installer environment, it's /var/log/syslog [11:59] http://paste.ubuntu.com/188151/ [11:59] is syslog [11:59] from save debug logs [12:04] so it certainly looks as though it's taken apt-setup/local1/repository into account [12:05] lots of errors about http://192.168.100.2/install/dists/hardy/ being busted in one way or another; maybe you should use a proper mirror for that bit? [12:05] 2894 Jun 4 10:25:22 in-target: Impossible de trouver le paquet Database [12:05] it's *very* unusual to have a package name begin with a capital letter [12:06] unheard of [12:06] in fact, policy explicitly forbids it [12:06] so that's probably why it breaks ... [12:06] "Package names must consist only of lower case letters (a-z), digits (0-9), plus (+) and minus (-) signs, and periods (.). They must be at least two characters long and must start with an alphanumeric character." [12:07] ok, I didn't know [12:07] I will change that [12:07] not certain that that's the problem, of course, but it's a good possibility [12:08] also, from an installer shell after the failed installation, 'cat /target/etc/apt/sources.list' to see if it really did add that second local repository [12:08] should be easy to see [12:09] yes, it's added [12:09] you're right [12:09] postgresql Database is a preseed file generation problem [12:10] it isn't even a package, king of noobish :) [12:10] kind of [12:14] I'm trying a new install [12:15] isn't there a bug with preseeding, my last character is never taken into account [12:16] as you might have seen, I always have to add a comment line [12:16] I think that's fixed in more recent versions [12:17] ok [12:18] what do you think about my firstboot method, is there a "better" way ? [12:19] I just link a script in /etc/rc2.d/ and the script mv the S entry to K entry as last command before reboot .... [12:21] as I said on #ubuntu-devel, I think it would be much better to arrange to install the packages before reboot. Firstboot arrangements are usually fragile and poor-quality in various ways. [12:22] For example, if you do it in the installer then you can easily (via 'in-target' etc.) take advantage of the installer's code to install packages non-interactively [12:25] for example, I need to confirm to my server that the install went right [12:26] how can I do that without firstboot ? [12:30] you have networking in the installation ... [12:31] anyway I don't care about every precise detail of your system, I'm mainly just saying that *installing packages* in a firstboot script is a bad idea [12:31] it's working inside preseed now, thank you for that, it was only Database which was set at the wring place [12:32] I know I have networking during installation, but as you say, it is during [12:32] thank you for your help and advices .. [12:34] well, you have a hook that runs pretty much at the end of installation too :-) [12:34] you might need firstboot if you aren't sure that the bootloader is going to work [12:35] might be stupid, but what do you mean with : hook that runs pretty much at the end of install .... [12:36] preseed/late_command [12:37] can't finish install fail ? [12:37] how sure can I be ? [12:38] in theory, but it's pretty unlikely [12:38] the bootloader is a far more likely failure point [12:38] if that's an issue for you, go ahead and carry on using firstboot for that, I won't stop you [12:38] ok, that's nice to know [12:38] using firstboot for that is a lot saner than using it for installing packages [12:39] you're right, installing packages inside firstboot isn't a very nice solution, especially when there're tools during install to do it [12:39] again, thank you very much for everything [13:28] I hope this isn't a pointless observation - I don't see anything immediately obvious in bugs.lp about it... I just did an install with the current daily karmic amd64 installer and my user is in no additional groups, so I can't sudo [13:28] Ng: can I get back to you after lunch? [13:28] (and it also looks as though /etc/sudoers is a default version, there's no group given sudo powers) [13:28] that sounds like user-setup crashed part-way through so I'll need to investigate the syslog [13:28] cjwatson: absolutely. I'll not touch it any further [13:29] thanks [13:29] back in an hour or two [13:30] ok [13:30] quick look suggests you are right, finish-install.d/06user-setup returned error code 1 after failing on some /dev/shm ecryptfs thing [13:30] I'll prune out my mount password and upload this stuff somewhere [13:30] if I can remember my lp password I'll file this all properly [14:07] (OOI, is that logging of the ecryptfs password to the installers syslog just there for development releases?) [15:31] Ng: *blink* the password is logged? poke kirkland ... [15:31] cjwatson: ? [15:31] Ng: any luck with the log? [15:32] 13:30 I'll prune out my mount password and upload this stuff somewhere [15:32] 13:30 if I can remember my lp password I'll file this all properly [15:32] 14:07 (OOI, is that logging of the ecryptfs password to the installers syslog just there for development releases?) [15:32] cjwatson: it shouldn't be logged to the installer log [15:32] cjwatson: but it does land briefly in a r------ file in /dev/shm [15:33] Ng: where is this logging of a password to syslog happening? [15:33] Ng: karmic encrypted home setup using the alt installer? [15:33] Ng: gimme some info and i'll go try to reproduce this [15:36] kirkland: yeah karmic amd64 alternate [15:37] just a sec... [15:37] http://mairukipa.tenshu.net/cmsj-installer.tgz [15:38] grep syslog for: zomghax [15:38] cjwatson: that tarball has all of /var/log/installer for your diagnostic purposes :) [15:38] kirkland: before I pruned it, it looked very much like the phrase I was encouraged to write down post-install [15:39] Ng: are you sure it wasn't just the signature? [15:39] addin [15:39] g auth tok with sig [0c34d8c80f951cdc] to the keyring [15:39] that's just a fingerprint for the much longer passphrase [15:39] Ng: I see those [15:39] Jun 4 11:58:26 user-setup: YOU SHOULD RECORD THIS MOUNT PASSPHRASE AND STORE IN A SAFE LOCATION: [15:39] Jun 4 11:58:26 user-setup: zomghax [15:39] Jun 4 11:58:26 user-setup: THIS WILL BE REQUIRED IF YOU NEED TO RECOVER YOUR DATA AT A LATER TIME. [15:39] Ng: ugh [15:39] perhaps this is fallout from whatever failed in user-setup [15:40] Ng: okay, let me track this down [15:40] Ng: i'm going to test jaunty first [15:40] if you or cjwatson want me to leave the install as-is, I will, otherwise I'll do a quick re-install without ecryptfs to get the laptop running ;) [15:40] cjwatson: tell this again, where do the install logs live post installation in a runnign system [15:40] Ng: would you mind holding onto it for a few minutes? [15:41] kirkland: sure, no problem. [15:41] Ng: this item has 100% of my attention at the moment [15:41] kirkland: that tarball was /var/log/installer/ on the target fs after installation [15:41] Ng: i lie, i'm pouring a cup of coffee first, then it has 100% of my attention :-) [15:41] hehe [15:41] fwiw, I used the current daily [15:42] so, yeah, ecryptfs just fucked here AFAICS [15:43] might be nice for user-setup-apply to recover and still do the sudo stuff [15:44] it seems like the resulting crypted home stuff works though [15:46] Ng: cjwatson: i have a successful karmic installation from last night, alt installer, encrypted home [15:47] Ng: cjwatson: i'm digging through /var/log/installer and I don't see any passphrase leakage on this system [15:47] oh, wait [15:47] there it is [15:47] shite [15:48] * kirkland tests jaunty, holding his breath [15:48] ecryptfs-setup-private seems to just lob it at stdout [15:49] anyway, you know more about this than I do, feel free to push a patch up when you have one [15:58] Ng: okay, i've reproduced the issue [15:58] Ng: you can have your machine back [15:59] cool :) [15:59] cjwatson: do you need anything else from the install? [16:00] no [16:01] ok, thanks guys :) [19:19] usb-creator: rgreening * r103 usb-creator/ (4 files in 2 dirs): [19:19] usb-creator: Added additional kde initialization bits required to startup a KApplication [19:19] usb-creator: (KAboutData, KCmdLineArgs, KApplication, etc). [19:19] usb-creator: kde_frontend is now launchable from usb-creator-kde, but still non [19:19] usb-creator: functional. [19:19] usb-creator: usb-creator-kde is essentially complete. [19:20] usb-creator: rgreening * r104 usb-creator/usbcreator/translate.py: Added translation class for use with KDE translation. [20:24] ubiquity: cjwatson * r2701 hardy-proposed/ (debian/changelog ubiquity/frontend/kde_ui.py): [20:24] ubiquity: * KDE frontend: [20:24] ubiquity: - Fix crash on selecting non-ASCII options in the "Use as:" menu while [20:24] ubiquity: creating a partition (LP: #247993). [20:25] ubiquity: cjwatson * r2702 hardy-proposed/ (debian/changelog ubiquity/frontend/kde_ui.py): Fix crash when entering a non-ASCII mount point (LP: #128554). [21:13] installation-report: cjwatson * r57 hardy-proposed/ (debian/changelog finish-install.d/94save-logs): [21:13] installation-report: Run gzip chrooted to /target rather than in the installer environment, [21:13] installation-report: since busybox-udeb isn't configured with gzip support, only gunzip [21:13] installation-report: (LP: #337276). [21:13] installation-report: cjwatson * r58 hardy-proposed/debian/changelog: releasing version 2.31ubuntu2 [21:33] usb-creator: rgreening * r105 usb-creator/bin/usb-creator-kde: [21:33] usb-creator: Refactor usb-creator-kde to use proper KDE cmd line options. [21:33] usb-creator: Re-order some internal bits (i.e move def's to top) and add a __main__. [21:33] usb-creator: Need to fixup kdesudo section or replace with policykit bits. [22:06] partman-basicfilesystems: cjwatson * r562 hardy-proposed/ (commit.d/format_swap debian/changelog): [22:06] partman-basicfilesystems: When formatting swap partitions, save the old UUID in a file rather than [22:06] partman-basicfilesystems: in a shell variable, as the latter approach eats NULs (LP: #336992). [22:11] evand: Yeah, I'm still having compilation issues. [22:11] evand: * runtime with future on Jaunty [23:38] ubiquity: cjwatson * r3266 ubiquity/debian/ (changelog intro-alpha.txt): Fix the intro message to have the correct release version and date. [23:42] ubiquity: cjwatson * r3267 ubiquity/ (d-i/manifest debian/changelog): [23:42] ubiquity: Automatic update of included source packages: base-installer [23:42] ubiquity: 1.99ubuntu2, debian-installer-utils 1.68ubuntu1, flash-kernel [23:42] ubiquity: 2.13ubuntu8, hw-detect 1.72ubuntu1, partman-base 129ubuntu8, [23:43] ubiquity: partman-target 59ubuntu2, silo-installer 1.15ubuntu2, user-setup [23:43] ubiquity: 1.26ubuntu1. [23:47] partman-ext3: cjwatson * r755 ubuntu/ (debian/changelog valid_filesystems/_numbers): Offer ext4 before ext3, since it's now the default. [23:48] evand: found a couple more bits that hardcoded ext3 :-/ [23:50] ubiquity: cjwatson * r3268 ubiquity/ (4 files in 3 dirs): [23:50] ubiquity: Various minor adjustments to cope with ext4 now being the default [23:50] ubiquity: filesystem. [23:51] partman-ext3: cjwatson * r756 ubuntu/debian/changelog: releasing version 56ubuntu3 [23:52] ubiquity: cjwatson * r3269 ubiquity/ (d-i/manifest debian/changelog): Automatic update of included source packages: partman-ext3 56ubuntu3. [23:53] pkgsel: cjwatson * r144 ubuntu/debian/changelog: releasing version 0.24ubuntu2 [23:56] ubiquity: cjwatson * r3270 ubiquity/ (debian/changelog scripts/install.py): Use block-attr rather than vol_id.