[00:51] Does anyone know how I can set a random hostname in preseed? [00:54] um [00:54] you could certainly do it in a script [01:26] ted_: where/how would I get the script to work with preseed? [01:29] brb [01:39] ok, sorry about that. Had to tweak my irssi and figured it was a good time to switch to freenode over ssl [11:13] debian-installer: cjwatson * r1299 ubuntu/ (8 files in 2 dirs): Remove multiarch (amd64-on-i386) syslinux configuration files. [12:15] debian-installer: cjwatson * r1300 ubuntu/ (build/util/syslinux-cfgs debian/changelog): [12:15] debian-installer: Don't install separate desktop syslinux configuration files; we don't [12:15] debian-installer: use them in Ubuntu. [12:21] debian-installer: cjwatson * r1301 ubuntu/debian/changelog: releasing version 20100211ubuntu1 [12:21] just, you know, for the avoidance of doubt: this is going to break [12:22] fun [12:34] ... and there we go, a build failure. odd, I tested on i386 [12:34] god, library reduction [12:34] * cjwatson prepares to dive down the rabbit-hole [12:55] hey guys has anyone else had issues with ubiquity on kubuntu lucid where you get an errorno5 input/output error on the live cd net install as well as the alternate cd? the ubuntu server installer works just fine. i would like to help the developers get this fixed in lucid as well as for maverick [12:59] input/output error almost invariably means a physical problem with the CD or the drive. [12:59] it's a very low-level problem and not an issue with the installer. [12:59] cjwatson: then can you explain to me why wiht ubuntu-server i dont have that issue [12:59] the fact that you see it on one image but not another is essentially coincidence; remember that the server installer uses a lot less of the CD than the other images do [13:00] ok [13:00] so it might be expected not to encounter as many problems [13:00] it's not impossible that it's a bug in the kernel's CD driver, but the probability is that it's a physical problem [13:01] cjwatson: even same issue with a live usb [13:01] im also starting to wonder if the capacity of my drive has something to do with it [13:01] I don't know what that might be [13:01] here's the definition of that error code from the libc manual: [13:01] -- Macro: int EIO [13:01] Input/output error; usually used for physical read or write errors. [13:02] it's basically never an application bug, anyway [13:02] cjwatson: whats funny is im not even using a cd to install lucid on this vm but an iso [13:02] could be a bug in the virtualiser, or an underlying I/O error on your hard disk that's being shown through to the VM [13:03] thing is i bet if i were to install ubuntu server that problem would disappear though [13:03] and the drive isnt too old [13:03] could it be a possibilty of bad ram [13:04] sorry, I've given all the help I can - if you're seeing unexplained EIO errors in a virtualisation environment then I suggest treating it as a possible bug in the virtualiser and going from there [13:04] i.e. working with people who know about the virtualisation system in question [13:05] cjwatson: thing is this happened to me on this desktop unvirtualized [13:05] frankly sounds like you have several different problems that are coming out with the same symptoms [13:05] but that's just a guess [13:05] all I can say is that there's basically no way EIO can be caused by an installer bug [13:06] ok [13:06] trying to eliminate causes of the issue [13:14] hmm, new eglibc has broken mklibs I think [13:26] ev: you make me blush :) ha! [13:27] just saw the usb-creator note change in LP [13:27] ty [13:38] well deserved [13:40] I try :) [13:41] now, If I could only get a paying job on foundations team :) [14:12] hello [14:39] thanks for your help yesterday, cjwatson. i managed to delude myself into thinking that i had added some modules to the initrd when i had in fact only added some other modules and then seen the ones in question appear because the installer feteched some stuff. once i put the right modules in the initrd, everything worked as expected. anyway, i have another installer question today. on 9.10 at least, it appears that some preseed configuration lines [14:39] that seem like they should affect how the installer tries to grab an address from DHCP don't do much. "d-i netcfg/dhcp_timeout string NUM" doesn't seem to change the number of seconds it takes for the installer to give up the first time it tries to get an address, nor does "d-i netcfg/dhcp_options select FOO" seem to cause any of the options from the window that appears after the first attempt to get an address fails to be selected automatically. i [14:39] 've checked /var/lib/cdebconf/questions.dat, and it looks like the values are getting into the debconf database, just not being used. does anyone know how this dhcp business is implemented or where else i could manually change the timeout value, etc.? [14:40] you're probably running into bug 315231 [14:40] Launchpad bug 315231 in netcfg (Debian) (and 1 other project) "netcfg/dhcp_timeout doesn't take effect when installing (heat: 2)" [Unknown,Fix released] https://launchpad.net/bugs/315231 [14:40] awesome [14:41] it works if you're trying to shorten the timeout, but not if you're trying to lengthen it beyond 60 [14:41] i can't get it to go past 30 [14:41] although [14:41] are you doing a netboot install? [14:41] yes [14:41] the network preseed file is fetched after netcfg runs, so you need to set those values on the kernel command line, not in your network preseed file [14:42] e.g. netcfg/dhcp_timeout=60 [14:42] oh cool [14:44] could i also add one like 'netcfg/dhcp_options="Retry network autoconfiguration"'? [14:50] oops xming froze [14:50] anyway [14:50] thanks for your help let's see if 60 seconds is enough [14:54] unfortunately not really, quoting doesn't work well on the kernel command line [14:54] that is not surprising [14:54] we should really turn those into short identifiers behind the scenes, so that you can just say netcfg/dhcp_options=Retry [14:54] or =retry [14:54] actually, looking at the code, you may find that netcfg/dhcp_options=Retry works right now [14:55] it's basically coincidence, but still [14:55] horrible horrible code [14:55] heh [14:55] make sure you capitalise it thus, it matters [14:55] whatever does the string matching only notices the first word anyway? [14:57] where does the code live? are you looking at the source for debian-installer or something? [15:01] netcfg [15:01] not debian-installer [15:05] 'apt-get source netcfg' [15:06] how can I put this, it doesn't do things the standard way for d-i [15:06] if (client->value[0] == 'R') { /* _R_etry ... or _R_econfigure ... */ [15:06] size_t len = strlen(client->value); [15:06] if (client->value[len - 1] == 'e') /* ... with DHCP hostnam_e_ */ [15:06] return REPLY_RETRY_WITH_HOSTNAME; [15:06] else if (client->value[len - 1] == 'k') /* ... wireless networ_k_ */ [15:06] return REPLY_RECONFIGURE_WIFI; [15:06] else [15:06] return REPLY_RETRY_AUTOCONFIG; [15:06] heh [15:06] } [15:06] * cjwatson passes the sickbag [15:07] that is pretty gnarly [15:07] but i mean if you have strlen why not use strcmp? [15:07] prejudices of the original author [15:08] the standard approach now is to have Choices-C with short identifiers like retry, retry-dhcp, ..., and to have human-readable translations of those per-language [15:08] okay [15:08] but I introduced the Choices-C mechanism some time after that code was written and nobody's got round to converting it over [15:08] that makes sense [15:08] one of these days ... [15:08] there's always something [15:09] i mean i can understand being afriad of strcmp, but strncmp is pretty safe, right? [15:09] it's just somebody being silly and over-obsessed by performance [15:10] it doesn't have a rational basis as far as I'm concerned :) [15:10] heh [15:10] but as it happens this prejudice is kind of convenient until we switch to Choices-C [15:31] dude that worked like nobody's business [15:31] one step closer to my hands-free build! [15:31] thanks [15:34] cool [18:43] partman-ext3: cjwatson * r766 ubuntu/ (77 files in 4 dirs): merge from Debian 59 [18:45] partman-ext3: cjwatson * r767 ubuntu/debian/changelog: releasing version 59ubuntu1 [18:48] cdrom-detect: cjwatson * r459 ubuntu/debian/ (cdrom-detect.templates changelog): Set cdrom/suite default to maverick; drop intrepid. [18:49] cdrom-detect: cjwatson * r460 ubuntu/debian/changelog: releasing version 1.32ubuntu4 [19:18] cjwatson, Did I ever submit my cdrom-detect/try-hdd patch to you? [22:40] cody-somerville: not sure [23:15] hi I've just been at #ubuntu.... I am having an issue setting up RAID 1 via a clean install -- the system basically won't boot properly after install [23:15] maybe someone is available to help? [23:50] can somebody help with ubuntu installation?