| === michael_e_brown_ is now known as michael_e_brown | ||
| === ebel_ is now known as ebel | ||
| CIA-24 | debian-installer: cjwatson * r894 ubuntu/ (9 files in 4 dirs): * Move to 2.6.24-12 kernels. | 13:57 |
|---|---|---|
| CIA-24 | debian-installer: cjwatson * r895 ubuntu/ (38 files in 2 dirs): * Update help text translations from Launchpad. | 14:04 |
| CIA-24 | debian-installer: cjwatson * r896 ubuntu/ (6 files in 2 dirs): | 14:08 |
| CIA-24 | debian-installer: * Remove translations for languages not supported by localechooser (and | 14:08 |
| CIA-24 | debian-installer: therefore unused by gfxboot-theme-ubuntu), specifically: ky mr ms oc ur. | 14:08 |
| CIA-24 | debian-installer: cjwatson * r897 ubuntu/debian/changelog: releasing version 20070308ubuntu33 | 14:15 |
| === ebel_ is now known as ebel | ||
| === cjwatson_ is now known as cjwatson | ||
| evand | cjwatson: thoughts on whitelisting /usr/local in ubiquity-preserve-home? This is in reference to bug 200582 . I'm on the fence as to whether or not it is a good idea. | 15:25 |
| ubotu | Launchpad bug 200582 in ubiquity "hardy alpha6 installer: "cleaning" /usr/local mount?!" [Undecided,New] https://launchpad.net/bugs/200582 | 15:48 |
| cjwatson | I definitely think that /usr/local should not be cleaned | 15:51 |
| cjwatson | wasn't that in the spec? | 15:51 |
| cjwatson | the reason is that the rest of the system officially doesn't care what's in /usr/local, so it shouldn't be necessary to clean it | 15:51 |
| cjwatson | also, further proof that a confirmation dialog is needed :) | 15:52 |
| evand | ok, no it wasn't, but perhaps I missed it when recording. | 15:52 |
| evand | indeed it is | 15:52 |
| mebrown | evand, are the latest fixes in today's daily cd? | 16:21 |
| evand | mebrown: yes, I tested them last night and they work for me, but if you run into any trouble please let me know. | 16:23 |
| mebrown | ok, will test. | 16:23 |
| mario_limonciell | evand, should the fixes for noninteractive be present in today's daily? | 17:14 |
| evand | mario_limonciell: which fixes? The issue when you preseed ubiquity/reboot is there, yes. | 17:20 |
| mario_limonciell | evand, okay i'll give it a shot then. thanks | 17:21 |
| evand | er the fix for the issue* | 17:23 |
| mario_limonciell | evand, hm i'm getting a handful of python exceptions thrown when trying. | 17:27 |
| mario_limonciell | I it's failing before partitioning even | 17:28 |
| mario_limonciell | and then spewing off "No root file system: No root file system is defined." "Please correct this from the partitioning menu." | 17:28 |
| evand | mario_limonciell: what method are you using to preseed? | 17:29 |
| mario_limonciell | via kernel command line | 17:30 |
| mario_limonciell | the same kernel command line that is used in automatic-ubiquity mode | 17:30 |
| mario_limonciell | but switching out automatic-ubiquity for noninteractive | 17:30 |
| evand | I mean are you using file= or url= ? Or are you setting all the debconf variables via the kernel command line? You need to have the installer fully preseeded. While it doesn't currently (I'm working on a proper fix in my local branch), any question it needs to ask should result in an error. | 17:31 |
| mario_limonciell | evand, i'm preseeding from the dell.seed file (not setting each of the variables via kernel command line). it is fully pre-seeded, no questions are asked in automatic ubiquity | 17:32 |
| evand | Also note that url= is broken, though I have a fix committed and pushed to trunk in casper. It's just a matter of uploading there. | 17:33 |
| mebrown | mario_limonciell, on todays daily? | 17:33 |
| mario_limonciell | mebrown, yes on today's daily | 17:33 |
| mebrown | mario_limonciell, on last week's daily (wed or so), it was getting all the way to the end | 17:33 |
| mebrown | i'm surprised that it is now failing so early | 17:33 |
| mario_limonciell | mebrown, you got a moment? I'll swing by and show you | 17:34 |
| mebrown | sure. | 17:34 |
| mario_limonciell | evand, and it is via file= | 17:34 |
| evand | hrm, can you post the dell seed by any chance? | 17:35 |
| mebrown | evand, have posted it in the past. Can post it again. | 17:36 |
| evand | I grepped through my logs and couldn't find it, but I imagine it's obfscuated by a pastebin url. | 17:37 |
| evand | oh actually | 17:37 |
| mebrown | how to kill the nointeractive installer? | 17:37 |
| evand | I have it right here | 17:37 |
| mebrown | it was a pastebin | 17:37 |
| evand | hrm actually, this one is pretty old. Would you mind pastebinning it again? | 17:38 |
| evand | ps aux | grep ubiquity should give you any processes associated with it. | 17:39 |
| mebrown | Saw a new and interesting message before it dies... something like "removing conflicting operating system files..." | 17:39 |
| mebrown | mario can send you the new dell seed | 17:40 |
| mebrown | and he is going to work on getting the logs to send | 17:40 |
| mario_limonciell | evand, mebrown explained how to gather the logs on this. I'll see what I can do about getting these to you. give me a few minutes | 17:40 |
| evand | mebrown, mario_limonciell: thanks | 17:41 |
| evand | re conflicting operating system files> That's clear_partitions in partman-target (https://wiki.ubuntu.com/UbiquityPreserveHome) | 17:42 |
| mebrown | evand, we dont have that, because it doesnt work for us. I manually clear the partitions in an early_command script | 17:43 |
| evand | It doesn't work for you? Can you elaborate? | 17:44 |
| mebrown | sure. The options were (iirc) something like | 17:46 |
| mebrown | 'remove all partitions' - wont work for obvious reasons | 17:46 |
| mebrown | 'remove all linux partitions' - also wont work because our data is on a linux partition | 17:46 |
| evand | ok, this is different. This is removing files from on a partition, not messing with the partition table. Basically, it removes /usr, /bin, etc from the mounted filesystems if the devices that they exist on are not marked to be formatted. This allows you to install Ubuntu over top of itself while preserving the home directory. | 17:48 |
| mebrown | evand, btw, https://wiki.ubuntu.com/UbiquityPreserveHome looks reasonable for us | 17:48 |
| mebrown | need to know how to preseed ---^ | 17:48 |
| mebrown | evand, well, it looks like the code is crashing right there | 17:49 |
| mario_limonciell | evand, http://paste.ubuntu.com/5541/ | 17:49 |
| evand | fantastic :/ | 17:49 |
| evand | actually, it's getting beyond that point, that's just the last part of partman_commit for the purposes of this log file. It's actually crashing with a bug I thought I fixed (the installer tries to communicate with debconf during the install when it already has the database locked by another process). | 17:52 |
| evand | When you paste your preseed I'll have a run with it and find out exactly why this is still occuring. | 17:52 |
| mario_limonciell | http://paste.ubuntu.com/5543/ | 17:53 |
| evand | thanks! | 17:54 |
| mario_limonciell | now if you'd like to emulate our partitioning layout, sda1 is a utility partition, sda2 FAT32, sda3 is ending up ext3, sda4 is ending up logical, sda5 is ending up swap | 17:56 |
| mario_limonciell | no prob on the seed file | 17:56 |
| evand | ok | 17:58 |
| mebrown | evand, stupid question: do you have tested sample preseed files published? | 18:18 |
| xivulon | shouldn't locale be en_US.UTF-8? | 18:20 |
| xivulon | I have filtered out the locales to use if that may help | 18:21 |
| xivulon | http://bazaar.launchpad.net/~ubuntu-installer/wubi/hardy/annotate/agostino.russo%40gmail.com-20080310010310-85389505oy1b51po?file_id=locales.ini-20080226233717-nxtkw1rax53kj76s-1 | 18:22 |
| evand | mebrown: I usually use a variant of http://evalicious.com/evan.seed, but that's definitely inaccurate in a few places as I grabbed it from a run of ubiquity. | 18:23 |
| mebrown | mario_limonciell, ---^ | 18:24 |
| mebrown | evand, you need to update yours too to use the replacement for "ubiquitypartman-auto/init_automatically_partitionselectGuided - use entire disk" | 18:25 |
| mebrown | and the choose_recipe as well | 18:25 |
| evand | indeed | 18:25 |
| mebrown | I had that on my list for after noninteractive is working | 18:25 |
| evand | mario_limonciell: on second look at the logs, you appear to be using an old CD. | 19:05 |
| mario_limonciell | evand, that should have been from today's or yesterday's daily. | 19:06 |
| evand | hrm | 19:06 |
| mario_limonciell | it was grabbed at 3:06 this morning | 19:06 |
| evand | what version of ubiquity is on it? | 19:07 |
| mario_limonciell | 1.7.12 - wow. that is out of date. | 19:07 |
| mario_limonciell | this *should* have been today's daily that was being seeded | 19:07 |
| mario_limonciell | let me double check with our factory image guys what happened. sorry for the mishap, likely fixed in "current" ubiquity :) | 19:08 |
| evand | no worries :) | 19:09 |
| mebrown | mario_limonciell, I have been manually updating casper/ and .disk/ from the daily because I sometimes run into these sort of problems... | 19:28 |
| mario_limonciell | mebrown, well hopefully we can find the source of how this happened. | 19:29 |
| mario_limonciell | evand, okay just got a fully noninteractive install to work properly | 21:22 |
| mario_limonciell | only thing i noticed that could be trouble is that it was 1000 percent (yes 1000 not 100) percent done before it got to 93 percent done | 21:22 |
| evand | mario_limonciell: indeed, I've noticed that as well and put it on my todo list. Do let me know if you guys run into any more trouble with the noninteractive frontend. | 21:27 |
| === cjwatson_ is now known as cjwatson | ||
| mario_limonciell | evand, okay. looks like there was a post install failure, but its on our end looking through the logs. | 21:30 |
| cr3 | I have a preseed file which contains the following for preseed/late_command: wget -O /target/tmp/late.sh http://host/path/to/late.sh; in-target sh /tmp/late.sh | 22:00 |
| cr3 | in late.sh, I have: echo "set /some/preseed/key value" | debconf-communicate; apt-get install -y my-package | 22:01 |
| cr3 | the first problem is that the /some/preseed/key is not replaced in my-package | 22:01 |
| cr3 | the second problem is that when I login to the machine after the installation, echo "get /some/preseed/key" | debconf-communicate only returns 0, meaning there is no value set | 22:02 |
| cr3 | is the reason that the debconf database is acting weird in /target during the installation? | 22:02 |
| cjwatson | probably because using in-target arranges for debconf passthrough, so that your debconf-communicate ends up going to the installer's cdebconf database | 22:02 |
| cjwatson | which isn't where you want it to go | 22:02 |
| cjwatson | lose in-target, use chroot /target instead, and add 'export DEBIAN_FRONTEND=noninteractive' before calling apt-get | 22:03 |
| cjwatson | I assume that you don't actually need interactivity, if you're preseeding it all | 22:03 |
| cr3 | absolutely, applying those changes now. thanks! | 22:03 |
| TheMuso | Now is it something with my setup, or do wubi installs on FAT32 always fail? Even when using the smallest amount of disk space that wubi offers to set up, every time I attempt to boot the install, fsck gives messsages about not being able to write to particular blocks, and segfaulting... | 22:05 |
| TheMuso | I also discovered that the host filesystem is mounted read-only... | 22:05 |
| TheMuso | Whereas on NTF,s all works fine, and the host is mounted read write. | 22:05 |
| === TheMuso_ is now known as TheMuso | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!