[00:11] could someone help me look at the error: http://paste.uni.cc/17831 [00:11] I try to upgrade from 7.04 - > 7.10 [00:12] I do upgrade 7.04 first, before upgrading [00:13] from these error message, most likely , it is eieio/emacs error [00:13] config error [00:13] h eieioow can I fix it? [00:13] how can I fix it? [00:13] might have more luck in #ubuntu-server [00:14] someone on #ubuntu sugguest me doing a freash install, that's thing I cannot afford to [00:14] yes, Flare183 [00:14] me? [00:14] yeah [00:14] not alot of people in here [00:14] but it's installer problem [00:14] yes [00:14] yeah but it's also a server problme [00:14] problem [00:14] twb is here [00:15] he is the person who help me when I was upgrading to 7.04 [00:15] twb:> dude lend a hand please [00:33] nothing [00:54] hello, should Iremove these error package first? [00:54] http://paste.uni.cc/17832 [01:05] thanks Flare183 [01:06] I fixed it [01:06] yeah! [01:06] great [01:06] how? [01:06] lol [01:06] yeah! [01:06] force remove eieio [01:07] so simple [01:07] just try it [01:07] oh wow [01:07] remove/eieio: purging byte-compiled files for emacs22 [01:08] i should of thought of that a long time ago but oh well now [01:08] 6 years since that error [01:08] great result! [01:08] dont need to reinstall, haha [02:17] d-i uses the wrong driver in xorg.conf (via instead of vesa). What is the most elegant way to force it to use a particular video driver? [02:18] Aha, found it: xserver-xorg xserver-xorg/config/device/driver select ati [05:58] cjwatson: ping? [05:58] That didn't work, it's still using the via driver :-( [08:09] twb: you need to ask the X people [08:14] Surely not the xorg people -- I'm pretty sure this is related the debconf code for xserver-xorg [08:29] which is maintained by the Debian/Ubuntu X maintainers [08:29] it's not part of the installer [08:29] (it happens to be fairly closely related, but we don't deal with it) [09:03] twb: which pci-id? [09:03] twb: do you have multiple video adapters installed? [09:06] cjwatson, i didn't see an ack on the branch to merge. Would you like me to file a bug against user-setup so it doesn't get forgotten? [09:16] superm1: oh, yes, please do file a bug, I haven't looked at the branch yet [09:16] ubiquity: cjwatson * r2389 ubiquity/ (d-i/Makefile d-i/update-control debian/changelog): [09:16] ubiquity: * Update d-i/update-control to work with the new Dpkg::Deps module in [09:16] ubiquity: dpkg-dev 1.14.8. [09:16] superm1: I'll probably actually just commit it upstream [09:17] cjwatson, okay cool. it's a small patch too. [09:17] superm1: I think it would be best to add an entry for it to the templates file too [09:17] even though it isn't strictly required if it's only going to be used by preseeding [09:19] yeah that's why i wasn't sure it was necessary [09:19] i'll add that, push, and then file a bug [09:19] superm1: I don't think the code you added for useradd will work, not that I hugely care [09:19] useradd -g requires an existing group [09:19] according to the man page the syntax looked right [09:19] probably needs to do groupadd separately [09:19] oh you're right [09:19] syntax is fine but semantics are wrong [09:19] :-) [09:21] the case that /usr/sbin/adduser isn't around, when does that come up for installations? [09:26] 19:29 (it happens to be fairly closely related, but we don't deal with it) [09:26] cjwatson: OK. [09:26] 20:03 twb: which pci-id? [09:27] tjaalton: I don't have the number offhand, but I spoke to the #unichrome people and they added it to their list of IDs [09:27] tjaalton: the default "via" driver works, but only on the external head [09:27] (It's a laptop, only one video card.) [09:27] Er, and "their" means openchrome.org, not unichrome.sf.net [09:28] tjaalton: here we go 17:49 twb: the ID is missing: (EE) VIA(0): Unknown Card-IDs (1071|8650); please report to [09:29] we probably should follow fedora and default to unichrome.. [09:29] ...so vesa works on both heads on gutsy, and svn openchrome works on gutsy+1, but via/gutsy only works on the external head. [09:30] twb, openchrome 0.3 is available in hardy [09:30] superm1: yeah, I didn't end up trying that. [09:30] superm1: it doesn't, in Ubuntu [09:31] superm1: it might do if you were producing a stripped-down Debian variant [09:31] cjwatson, all i was saying is that I packaged up openchrome for hardy a few weeks ago [09:32] cjwatson: should I be hassling the Debian X Strike Force, or some Ubuntu derivate thereof? Either way, do you know if they have a dedicated channel (maybe on OFTC)? [09:32] twb: #ubuntu-x [09:32] Thanks. [09:33] superm1: I was replying to your question about adduser [09:33] cjwatson, oh :) [10:23] debian-installer: cjwatson * r862 ubuntu/ (build/util/help-to-gfxboot.py debian/changelog): [10:23] debian-installer: * Encapsulate UTF-8 encoding pain differently in [10:23] debian-installer: build/util/help-to-gfxboot.py. (The resulting code is longer but I think [10:23] debian-installer: the horribleness is better-positioned.) [10:24] excellent, my CIA bot cunning worked [10:24] debian-installer svn commits -> #debian-boot, bzr commits -> #ubuntu-installer [10:24] logic not perfect but it'll do for now [10:25] not sure I feel like manually writing out all the d-i subprojects though ... maybe some other day [10:26] oem-config: cjwatson * r389 oem-config/ (d-i/Makefile d-i/update-control debian/changelog): [10:26] oem-config: * Update d-i/update-control to work with the new Dpkg::Deps module in [10:26] oem-config: dpkg-dev 1.14.8. [10:34] ok, automation works wonders, I think I have all the d-i components set up for CIA notifications now [10:35] at least all the ones with bzr imports that I know about === cjwatson_ is now known as cjwatson [12:11] ubiquity: cjwatson * r2390 ubiquity/ (d-i/manifest debian/changelog): [12:11] ubiquity: * Automatic update of included source packages: apt-setup 1:0.31ubuntu2, [12:11] ubiquity: base-installer 1.86ubuntu1, debian-installer-utils 1.50ubuntu1, [12:11] ubiquity: partman-auto 73ubuntu1, partman-base 114ubuntu1, partman-basicmethods [12:11] ubiquity: 36, partman-efi 14ubuntu1, partman-partitioning 54ubuntu1. [12:19] ubiquity: cjwatson * r2391 ubiquity/debian/po/ (79 files): debconf-updatepo [12:59] ubiquity: cjwatson * r2392 ubiquity/ (4 files in 3 dirs): [12:59] ubiquity: * Update partman extensions to cope with changes in partman-base 114 [12:59] ubiquity: (/lib/partman/definitions.sh -> /lib/partman/lib/base.sh). [13:02] ubiquity: cjwatson * r2393 ubiquity/debian/ (80 files in 2 dirs): mark ${PARTITIONS} untranslatable [13:04] ubiquity: cjwatson * r2394 ubiquity/debian/ (80 files in 2 dirs): sync templates with migration-assistant changes [13:11] ubiquity: cjwatson * r2395 ubiquity/debian/changelog: releasing version 1.7.2 [13:24] Sheesh. Calm down. [13:26] cjwatson: I'm supposed to be implementing some form of iscsi support in the installer. Can you think of any other installer components doing something like that that I could look at for inspiration? AFAIK (I'm a complete iscsi n00b), it's mostly a matter of passing some information to the iscsi initiator thing which (AFAIK) creates some block devices for the installer to use.. [13:27] err, sounds a little like how dmraid works, though that's a freaky hard [13:27] er, a freaky hack [13:28] or maybe the other lvm/raid stuff [13:28] Hmm.. [13:28] I was thinking that it should preceed all of that. [13:28] All of the lvm/raid stuff needs to happen at the same time, because it can be layered like crazy. [13:29] that's not the case for iscsi. It just makes some block devices available for the rest of the partitioning magic to use. [13:32] you could use an init.d script then [13:32] /lib/partman/init.d that is [13:33] Sounds like a sensible place for it. I can interact with the user there? [13:33] debconf style, of course. [13:33] in principle yes, though I don't think anything else in init.d does [13:34] hmm [13:34] from the sound of it, a better place would actually be disk-detect (in the hw-detect source package) [13:35] since it's fundamentally just a weird case of disk block device detection [13:35] It is. [13:36] And yes, that does sound more sensible. [13:36] I'm stilling trying to get an overview of all the components and how they fit together, so I'll probably be bouncing this sort of thing off of you for a bit. Hope that's ok. [13:36] no problem at all [13:37] I know it's a bit of a maze at first [13:37] Think of it as an investment :) [13:37] I do :) [14:07] cjwatson: What's this doing at the top of disk-detect.sh? [14:07] if [ "$(uname)" != Linux ]; then exit 0 [14:07] fi [14:07] Seems a bit superfluous :) [14:07] Ah, hurd? [14:09] I imagine so [14:09] or freebsd [14:10] d-i doesn't actually work completely on either of those yet, but has been partially ported in places [14:10] might as well leave the stuff in since it's generally cheap and saves doing it again when the time comes [14:11] Sure. I didn't think of the freebsd or hurd use cases and then it seemed a bit odd. [14:38] cjwatson: I'm trying to work out why disk-detect get installed and called.. Any hints? [14:41] cjwatson: I can't find it anywhere in any dependency chain.. d-i itself only mentions it in documention, afaics.. [14:42] it's got an Installer-Menu-Item control field, so main-menu calls it [14:43] it goes roughly in menu-item number order, modified by requirements imposed by dependencies [14:51] Ah, so it does. [14:51] What makes sure that it get installed at all, though? [15:04] Ah, think I got it. [15:05] grep is hard.. [15:07] soren: partman-base is Priority: standard, depends on harddrive-detection, provided by disk-detect [15:07] anything Priority: standard gets installed by default [15:08] cjwatson: Oh. disk-detect is also mentioned in the pkg-list of debian-installer. [15:08] Ah, but that just includes it in the installer, that doesn't actually install it. [15:09] right [15:09] Ok.. Got it. [15:09] s/installer/initrd/ [15:09] Er, yes. [15:10] Ok, so all the ordering stuff mentioned here http://d-i.alioth.debian.org/doc/talks/debconf6/paper/ is generally dictated by Installer-Menu-Item. [15:39] cjwatson: I'm just looking at the code to support dmraid.. It seems to me that the only way to enable it is by preseeding.. Is that a general approach to these more exotic kinds of things, or is it merely because its considered experimental? [15:42] probably best not to use dmraid as an example [15:45] cjwatson: Oh? [15:45] it's a nasty hack in a bunch of ways, one of which is the UI issue you mention [15:46] if you can detect whether it's sane to enable iscsi, that's a lot better [15:46] but if you can't detect it, then yeah, you're probably going to have to either (a) use preseeding (b) make it an optional installer components [15:46] s/s$// [15:47] cjwatson: If I go with (b) I need to create a separate udeb, I suppose.. How would I then get access to that in the installer? [15:47] Is there a way to get a list of the available udebs and the choose to install one? [15:48] in expert mode it'll ask [15:48] or you can preseed anna/choose_modules=some_extra_thing [15:48] Oh, I see. [15:49] if it's not very much code, it would probably be easier to put it in disk-detect and have a preseed for that [15:49] we could even alias it ... iscsi => hw-detect/enable_iscsi or whatever [15:50] I don't think it's going to be a lot of code, no, but I need to create a udeb anyway to have the userspace tools available in the installer. [15:50] I didn't think of that until just now. [15:50] right, but that's a udeb built out of the main iscsi source package or whatever; installer integration is often separate [15:50] though, hmm, it would have to depend on the tools, that does suggest a separate udeb [15:51] Well, the dmraid code just anna-install's the dmraid-udeb if it gets activated. [15:53] I can't really decide which approach makes more sense. [15:53] Doing the configuration after partman is kind of useless, I suppose, and the only way to make sure it's done prior to that is to hook it into disk-detect (or something simiar). [15:54] Or? [15:55] And from there conditionally install the userspace utilities using anna if the user needs iscsi. [15:58] , sorry [16:38] I'm in a meeting myself, so no worries. [16:50] ok, off [16:50] you definitely want to do it before partman [16:51] using anna-install would be OK [16:51] that way you wouldn't need a separate udeb for the installer integration, and could just use anna-install instead of a hard dependency [17:00] Exactly. [20:06] Can I get my server running apt-cache to handle the security updates too? Right now the clients are looking out to the internet to download security updates and the firewall is blocking them [20:07] Right now I'm using: d-i mirror/http/directory string /us.archive.ubuntu.com/ubuntu and d-i apt-setup/security_host us.archive.ubuntu.com because install wasn't working without it. Now I opened the firewall just for that client and it goes through [20:07] I'd rather it get it all from the server though [20:31] Goosemoose: Why not just point it at your apt-cacher? [20:36] for the directory string? [20:36] i had tried that a few days ago and got an error, maybe it was just a fluke, i can try it again [20:36] right now i have d-i mirror/http/hostname string 10.0.2.131:3142 [20:37] Goosemoose: No, for the security_host. [20:38] so you're saying add d-i mirror/http/directory string 10.0.2.131:3142 [20:38] No. [20:38] Leave mirror/blah as is. [20:38] Do: [20:39] di apt-setup/security_host string 10.0.2.131:3142 [20:39] Er.. [20:39] d-i apt-setup/security_host string 10.0.2.131:3142 [20:39] Hm.. Hang on. [20:40] Oh, I see the problem now! [20:40] Hah.. [20:41] I'm listening :D [20:42] I said I saw the problem. Not the solution :) [20:42] It seems that /security is hardcoded. [20:42] haha [20:42] Er.. [20:42] doh [20:42] I mean /ubuntu [20:42] ok, that would explain why i couldn't get it to work [20:43] soooo, should i create a symlink called ubuntu or something to trick it? [20:43] That won't work. [20:43] apt-cacher doesn't look at file system paths, it just uses the first bit of the URI to know which host to actually ask for the files. [20:44] Hm... I belive apt-cacher has some sort of aliasing system.. [20:44] It does. [20:44] You could add something like: [20:44] so even if i type in d-i apt-setup/security_host string 10.0.2.131:3142 its still looking at the /ubuntu location [20:45] ok, listening [20:45] path_map = ubuntu security.ubuntu.com/ubuntu [20:45] to apt-cacher.conf [20:45] ok [20:45] you said to leave d-i mirror/http/directory string /us.archive.ubuntu.com/ubuntu alone? [20:45] just change the security host [20:45] As long as you make sure you always remember to use "/nn.archive.ubuntu.com/ubuntu" when you're not trying to get to the security stuff. [20:46] Goosemoose: Yes. [20:46] ok, but then wouldn't the client need access to that site? [20:46] Ok, from the top: [20:46] d-i mirror/http/directory string /us.archive.ubuntu.com/ubuntu [20:47] d-i mirror/http/hostname string 10.0.2.131:3142 [20:47] d-i apt-setup/security_host string 10.0.2.131:3142 [20:47] Those three lines will result in the following in /etc/apt/sources.list: [20:48] deb http://10.0.2.131:3142/us.archive.ubuntu.com/ubuntu gutsy main restricted blahblahbhal # This is the regular (non-security) archive [20:48] deb http://10.0.2.131:3142/ubuntu gutsy main restricted blahblahbhal # This is the security archive [20:48] The reason the latter works is due to the mapping we defined in apt-cacher.conf [20:48] i.e. path_map = ubuntu security.ubuntu.com/ubuntu [20:48] ok i see [20:48] thanks, ill try that now [20:49] What I was just pointing out was that it's an easy mistake to just put "deb http://10.0.2.131:3142/ubuntu gutsy main restricted blahblahbhal" in your sources.list if you're writing it by hand, but that will only get you the security updates, not the regular (non-security) archive. [20:50] gotcha [20:50] ...so you just need to keep that in mind.. Now that I've mentioned it, it might spring to mind if you start seeing a lot of stuff missing from your archive :) [20:50] lol [21:08] there's a bug about the hardcoding of /ubuntu, it just needs to be cleaned up in a couple of places, and ideally the preseed template should go to Debian so that we don't end up diverging on its name [21:10] cjwatson: You mentioned something about debconf aliases.. [21:10] cjwatson: Does that just provide a means for defining short-hands for keys or is it for key *and* value? [21:11] cjwatson: I.e. can we make "iscsi" on the kernel command line mean "disk-setup/iscsi/enable=true"? [21:11] preseed_aliases in the preseed package [21:11] no, I think it would have to be iscsi=true [21:11] but I think that would be clearer anyway [21:11] Oh, definitely. [21:12] The other thing would just be a bonus. [21:12] Ah.. [21:12] No. [21:12] :) [21:12] Never mind. [21:14] debian-installer: cjwatson * r863 ubuntu/debian/changelog: releasing version 20070308ubuntu23 [21:14] xivulon: yes, that bug list is fine [21:15] m,kkkkkkkkkkkkkkkkkkk[-0-;plkl-;[,0m [21:16] same to you [21:16] kids at the keyboard, [21:16] heh [21:16] sorry [21:16] evand: well done on -core-dev! [21:17] cjwatson: The code that grabs the stuff from the kernel commandline and turns it into preseed stuff... Where's that? greping through the preseed code for /proc/commandline doesn't give me anything. [21:17] it's spelled /proc/cmdline ... [21:17] but in any case, it actually shows up as environment variables [21:18] *headdesk* [21:18] /bin/env2debconf, called from /lib/debian-installer-startup.d/S30env-preseed [21:18] Er... say what? [21:18] How does it end up as environment variables? [21:19] I believe the kernel puts stuff from the kernel command line in the environment by default [21:20] I've never noticed that. How odd. [21:20] it's possible that other init implementations clear it out [21:20] but busybox init doesn;t [21:20] doesn't [21:21] That doesn't add up... [21:21] The init in our initramfs goes through the contents of /proc/cmdline to fish them out.. [21:22] Why would it do that if they're already there? Hmm.. [21:22] don't recall, I just know it works. :) [21:23] I'm just trying to work out if there was any simple way to magically turn "iscsi" on the kernel command line into something useful. That would really be optimal. [21:24] Even if it just set it to an empty string would be usable. [21:24] I don't think that's a good idea; it's too likely to clash with future options parsed by the kernel itself [21:24] even iscsi=true is skating on thin ice there [21:25] all the foo/bar=blah things are safe because the kernel never uses keys containing / [21:25] Hm... point. [21:30] thanks cjwatson ! [21:31] \o/ [22:19] damnit, still getting installation step failed and select and install software [22:19] if i go back to the menu and click install it goes fine [22:22] soren, the client still tried to access the internet via http://91.189.88.31/ubuntu/dists/gusty/Release [22:27] I can't imagine why.. cjwatson is likely to have better guesses than me. [22:36] cj, you still around? [22:36] as soon as the install finishes here i can get the install log [22:39] Goosemoose: That would be lovely. [22:42] almost there [23:21] ok, got the file [23:21] posting it to pastebin now [23:27] wow its 6 megs [23:27] guess ill need to post it to my site [23:30] oops only 600k [23:30] was gonna say! [23:32] ok its up http://www.damien-hs.edu/syslog [23:33] soren or cjwatson, if you guys see the problem ,please let me know [23:33] Dec 18 22:18:48 in-target: E: Method http has died unexpectedly! [23:33] Goosemoose: How are you running your apt-cacher? [23:34] looks like random network interruption, hard to say for sure [23:34] at any rate that's internal to apt [23:34] hmm, the server and client are hooked up to the same switch [23:34] could be a bug in the server, sure [23:34] soren, followed that apt-cacher ubuntu install guide [23:34] mvo would be the best person to debug that [23:34] if it's possible to give him a reasonably compact reproduction recipe [23:35] funny thing is if i go back when i get the software install errror to the menu, then click install software it goes through [23:35] right, hence my "random" comment [23:35] it might well work fine if you ran it through again [23:35] happens every time though [23:35] no, ran it 5 times [23:35] odd [23:35] i still see it attempting to go through the firewall too, though I think it's sucessfull, but I want to avoid that for the other clients. This one I gave all access to [23:36] It's apt-cacher that's messing up. [23:36] interesting [23:36] Sorry, I lost my network for a little bit. [23:36] hmm [23:36] Dec 18 22:11:58 base-installer: 500 Can't connect to ubuntu:80 (Bad hostname 'ubuntu') [23:36] no problem soren [23:36] ok, there is no machine called ubuntu:80 [23:36] Did you set those mappings as I said? [23:36] so why would it be connecting to that? [23:36] yes [23:36] it dies twice, so it may just not be reliable enough to last for a full install ... [23:36] Goosemoose: It's apt-cacher that's trying to connect to at host called ubuntu. [23:37] strange [23:37] Goosemoose: because those mappings aren't taking effect. [23:37] path_map = ubuntu security.ubuntu.com/ubuntu [23:37] i have that [23:37] let me pastebin my apt config [23:37] Where's that guide you mentioned? [23:37] oh yes, I misread that as a client-side error, good catch soren [23:37] i have it printed, let me find it online [23:37] I don't remember if it runs as a mod_perl thing.. [23:38] http://www.google.com/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Fubuntu-tutorials.com%2F2007%2F01%2F08%2Fsave-bandwidth-during-updates-with-apt-cacher-ubuntu-610%2F&ei=0VloR7-WOaXqpATM9-nWBA&usg=AFQjCNEK1wayl7Xg7LchBYrDNfOg910M9Q&sig2=Y0hR-ArQXrHEw20kO12sIQ [23:38] If it does, you might need to restart apache for those changes to take effect. [23:38] * cjwatson knows nothing about apt-cacher and will go to bed [23:38] hmm, good point, it does run off of apache, let me try restarting apache [23:38] good night cjwatson, thanks for the help [23:38] Goosemoose: BTW, if you want my attention, you need to say "cjwatson" for it to effectively highlight; I don't highlight just on "cj" because it would be a bit crazy [23:38] sure thing [23:38] cjwatson: When you've done that see what happens if you go to: http://10.0.2.131:3142/ubuntu [23:38] I'm on enough channels that I need mechanical assistance to keep up ;) [23:39] soren: not me :) [23:39] let me see if i can go to that before i reboot [23:39] cjwatson: Gah.. [23:39] yeah i have mine on my mirc at home [23:39] Goosemoose: When you've done that see what happens if you go to: http://10.0.2.131:3142/ubuntu [23:39] i can see it now before i reboot apache soren [23:39] What can you see? [23:39] looks like a php info type page [23:40] Apt-cacher version 0.1 [23:40] has the config settings [23:40] Usage: edit your /etc/apt/sources.list so all your HTTP sources are prepended with the address of your apt-cacher machine and the port, like this: [23:40] deb http://ftp.au.debian.org/debian unstable main contrib non-free [23:40] becomes [23:40] deb http://yourcache.example.com:3142/ftp.au.debian.org/debian unstable main contrib non-free [23:40] Does it show the mapping? [23:40] no [23:40] so let me reboot and then see if it does [23:42] Try this: http://10.0.2.131:3142/ubuntu/dists/gutsy/Release [23:43] ok restarted [23:44] still dont see the mapping [23:44] and that url never loads [23:44] Odd. It works for me. I've just set the same mapping. [23:44] Is your apt-cacher machine allowed to access security.ubuntu.com? [23:44] hmm [23:44] Firewallwise? [23:44] yes, it has full access [23:45] yup, looks like it's in new zealand when i ping it [23:45] Did you stop and start apache? Or just reload? [23:46] restart [23:46] Goosemoose: That's very odd. [23:46] ill try stop/start, should be the same [23:46] Goosemoose: Mine shows it just fine on that page. [23:47] http://pastebin.com/d7a66ab5 [23:47] there's my apt conf file [23:48] Looks about right. [23:48] hmm [23:48] i dont have to restart apt right? [23:48]