[00:02] cjwatson: okay, cool, thanks for the info [10:01] davmor2: hi [10:01] hello xivulon [10:02] I believe you wanted to know about the status of wubi [10:02] xivulon: yes the python port [10:03] the current version should go through the installation, with a few quirks in the GUI (no transparency of labels/bitmaps, python subprocess calls make a shell pop out...) [10:03] but it can and should be tested [10:03] that already includes umenu [10:04] xivulon: so the current version of wubi should be the python version then yes? [10:04] I think so [10:05] which version is python and I'll check it against the cd [10:06] here is my todo list at the moment, http://paste.ubuntu.com/110210/ [10:06] will create lp bugs at some stage [10:07] python gets embedded into the executable when you build it [10:07] if you get the source, run make, and accept all the default options from the windows that popup [10:08] bzr lp:wubi && cd wubi && make [10:08] evand, what is the status of the wubi migration support in ubiquity? [10:09] I don't think I'll have time to work on that this cycle. [10:09] :( [10:13] apologies, but I have a lot of ground to cover with ubiquity this cycle [10:17] evand np, I am in the same boat for different reasons [10:18] not sure if the load could be split into subsections, I might be able to help in one of them [10:20] At this point I wish I had a clue about programming and development. But don't :( [10:44] debian-installer: cjwatson * r1018 ubuntu/ (build/config/lpia.cfg debian/changelog): Move lpia to 2.6.28-1 kernels. [10:47] debian-installer: cjwatson * r1019 ubuntu/ (4 files in 2 dirs): Move mainline architectures to 2.6.28-5 kernels. [10:55] lool: so, this nslu2 d-i build failure. Do you know if there's actually a limit on the size of the resulting image (at least one I'm likely to run into)? Can I just increase the size we pad the kernel to so that it's the next block size up from our kernel's size? [10:55] lool: or do I need to find a way to decrease something else in parallel? [11:02] cjwatson: I think there's a limit of kernel + initramfs size, but the individual limits for initramfs and kernel aren't matching our requirements [11:02] cjwatson: I don't have a NSLU2, but NCommander has one; he could help you clarify [11:03] cjwatson: My understanding is that because this image is only used to run the installer, we don't really care in keeping the same size allocated for kernel versus initramfs [11:03] cjwatson: Do you happen to know how the flash is formatted in the end (during install)? [11:04] no idea [11:04] it may not be supported properly at all [11:04] why don't I just bump the block size, I don't really care that much [11:04] if it doesn't work we can poke at it later [11:05] cjwatson: Hmm ok; I think you will spend as much time also decreasing the size in blocks of the other part [11:05] As to keep the whole size constant [11:06] cjwatson: Do whatever you think can get us an image to build and I'll ask NCommander whether he has time to test the image [11:07] debian-installer: cjwatson * r1020 ubuntu/ (build/config/armel/ixp4xx/netboot.cfg debian/changelog): [11:07] debian-installer: Pad the nslu2 kernel to 16 blocks rather than 11, so that the Ubuntu [11:07] debian-installer: kernel fits. [11:14] debian-installer: cjwatson * r1021 ubuntu/debian/changelog: releasing version 20081029ubuntu10 [11:23] Thanks === davmor2 is now known as davmor2-lunch [14:00] regarding the new time zone map === davmor2-lunch is now known as davmor2 [14:01] given that there are 40 time zones, would it be ok to just show the major 12 [14:01] and leave the rest to the drop down boxes [14:01] cjwatson: thoughts? [14:02] evand: I think it might get confusing. [14:02] davmor2: how so? [14:05] evand: Having people be able to see their zone on the map is important. I think if it isn't there and obvious it might throw people. Plus you may alienate people who aren't in the zones that got selected. [14:07] davmor2: I'm not entirely sure I agree. People are used to selecting their time zone from a numerically sorted list from the web or from using Windows, and I suspect those people in Iran or on small islands are probably used to being second classed when it comes to this. [14:08] hmmm I suppose is there a major gain from dropping them though? [14:08] The alternative would be to either go back to selecting cities on the map, or allowing them to scroll through all 40 timezones (feasible, just more work for kwwii) [14:10] http://upload.wikimedia.org/wikipedia/commons/e/e7/Timezones2008.png gives a pretty good picture of the problem [14:12] evand: Point taken trying to get that lot into a small map would look kinda ugly [14:13] scroll was probably the wrong word as well. They'd move the mouse over, or use arrow keys to cycle through the zones. [14:16] evand: Oh that makes more sense. So you hover over the major ones and it lists the minor in a similar way to a tag on a photo kinda thing. [14:17] no [14:17] you move the mouse over the timezones, it highlights a zone, you click on it and it sets that as your zone [14:18] if you want to refine further, say select a different city, you use the rightmost drop down box of the two that sit below the timezone map [14:19] Ah right with you. [14:19] you can also navigate via keyboard on the map [14:19] which is what I was getting at when I said use the arrow keys to cycle through the zones [14:21] cjwatson: I notice sometimes i get stopped during the installer on my kickstart with a 'unable to automatically remove LVM data' error, is there any way to just force this to go ahead and do it? [14:22] evand: Ah okay that's not so bad. [14:25] cjwatson: or at least if you do get stuck in a automatic/kickstart install you should be able to manually partition [14:38] cjwatson: what I mean is, create an LVM that spans two physical disks, say sda and sdb, then run my kickstart, it stops and says that it can't remove the LVM because it spans two physical disks, then it wont let you manually partition it either, essentially the installer breaks. [14:40] I think you can work around that by adding both disks to partman-auto/disk in a preseed. Not sure how to do that in kickstart, and definitely not confident about that solution. [14:42] right, but what if not all of your machines have 2 disks. [14:43] there should be a way to just universally wipe stuff out [14:45] the problem really is that redhat and centos's default is to span LVM to both disks during a kickstart, so if we take a machine that once has redhat/centos on it and then try to kickstart onto ubuntu its going to fail every time [14:47] would be good if at least if it dies because it notices a physically spanned LVM that it lets you go to the partitioner manually. [14:47] instead of just saying 'there is no root partition' over and over [14:50] actually let me see what our rhel 5 install does it shouldnt span the two drives automatically [14:56] perhaps we should make clearpart --all use the documented partman-lvm preseeding for that [14:56] (see the installation-guide, you can set it manually) [15:06] cjwatson: what I mean is, if you are using kickstart to automatically set up a machine and you hit that LVM wall where you have a LVM spanned on two disks, the installer breaks, it doesn't let you then manually partition or anything. [15:06] ugh, cdimage is intolerably slow today [15:07] actually, given my proximity, I must be doing something wrong [15:09] DogWater: yes, I understood [15:09] DogWater: please do file a bug on kickseed, though, I won't remember an IRC discussion :) [15:12] cjwatson: and to your knowledge is there a way to force it to remove the LVM from a multi physical disk span? [15:12] via kickstart or preseed? [15:12] i really need it to work no matter what [15:16] did you try the one documented in the installation guide? [15:16] # If one of the disks that are going to be automatically partitioned [15:16] # contains an old LVM configuration, the user will normally receive a [15:16] # warning. This can be preseeded away... [15:16] d-i partman-lvm/device_remove_lvm boolean true [15:17] i just downloaded today's daily iso ... it didn't pick up ecryptfs-utils-69-0ubuntu1 update ... anything i need to do to get that in the next build, or is it automatic? [15:17] it's automatic [15:18] cjwatson: ah jeesh sir that doesn't work if there is an LVM partition spread across multiple physical disks [15:18] there were live CD problems today [15:18] DogWater: I see; I just wanted to make sure you had actually tried that before continuing [15:18] i.e if you have a LVM on sda and sdb [15:18] DogWater: please file a bug with the details; I can't usually diagnose this sort of thing off-the-cuff on IRC [15:18] cjwatson: gotcha, thanks. [15:20] cjwatson: is there a second spin coming today? [15:28] do you have a link to the correct place to file the bug handy im not able to find the bookmark i made [15:29] kirkland: don't know, sorry, in a TB meeting [15:29] DogWater: https://bugs.launchpad.net/ubuntu/+source/kickseed/+filebug [15:30] cjwatson: congrads on the TB, btw :-) [15:34] ta [16:22] cjwatson: Seems you really had to give that flash back to some other part for NSLU2 ;-) [16:23] lool: why don't I just turn off nslu2 [16:23] for the time being [16:23] cjwatson: Ok; otherwise I just had confirmation from Michael that he had a NSLU2 and was willing to test [16:23] do we care about it? [16:23] cjwatson: Well it's a popular community device [16:23] I need a patch more than I need testing :) [16:23] but for the time being, I'd settle for d-i building [16:23] cjwatson: Ok; I'll tell him to have a look then [16:24] cjwatson: You're the boss [16:27] hey cjwatson [16:28] hi [16:28] cjwatson, I'm looking at your NSLU2 d-i woes, and I was hoping I could give you a hand (if you needed it) [16:28] debian-installer: cjwatson * r1022 ubuntu/ (build/config/armel/ixp4xx/netboot.cfg debian/changelog): Disable nslu2 image until somebody gets it to build. [16:28] sure [16:28] please send a patch [16:28] It looks like we need to kernel the kernel to be "less fat" [16:29] oh, hang on, do we? [16:29] Padding ./tmp/ixp4xx_netboot/initrd.gz.nslu2 from 3369710 to 6291440... padding with 2921730 bytes [16:29] we're padding that massively excessively, far more than we needed to grow the kernel partition [16:29] Oh, you adjusted the paritions? [16:29] yes [16:29] * NCommander hopes that leaves enough room for d-i [16:29] d-i is in that initrd already [16:30] And the whole thing is less than 8MB? Cool :-) [16:30] I can look at testing it once images are built (or I can build my own if need be) [16:30] NCommander: I think you should be able to reproduce by rebuilding d-i on babbage and adjusting the pad / sizes to build [16:31] NCommander: The idea would be to make debian-installer build so that we get images [16:31] Works for me. [16:31] I'm changing the initrd pad size to 5636080 [16:33] debian-installer: cjwatson * r1023 ubuntu/ (build/config/armel/ixp4xx/netboot.cfg debian/changelog): revert r1022, I think I see a fix [16:37] debian-installer: cjwatson * r1024 ubuntu/ (build/config/armel/ixp4xx/netboot.cfg debian/changelog): [16:37] debian-installer: Reduce nslu2 initrd padding by the same amount as I previously increased [16:37] debian-installer: the kernel padding, since there's plenty of room for the initrd. [16:37] NCommander: please test r1024 [16:38] as soon as bazaar truck finishes checking out, I'll kick off a build [16:38] * NCommander has been told the ixp4xx kernel should in theory just work so we'll find out soon enough :-) [16:45] evand: what's happening with the encrypted-home fix? [16:47] cjwatson: should be coming together for tomorrow. I'm in the middle of the blacklist fix, and kirkland has apparently fixed the kernel module / ecryptfs-setup-private interaction in the latest upload. [16:47] debian-installer: cjwatson * r1025 ubuntu/ (4 files in 2 dirs): Move mainline architectures to 2.6.28-6 kernels. [16:47] evand: yeah, that blah blah about user= invalid option is fixed [17:11] cjwatson, maybe its just me, or LP, but it might be awhile before I can test d-i, its STILL checking out, and the bar is barely moving past 25-26% [17:13] it's in an old branch format, I probably need to upgrade it [17:13] but d-i svn upstream used to contain the manual as well [17:13] and unfortunately you still have to download all of that history even though the files have been removed [17:13] you could try bzr checkout --lightweight [17:34] thanks, that helped [17:37] Hrm, getting rid of "Guided - use..." is not going to be as easy as I thought, given the lack of " - " in some of the po files. Looks like we'll have to have additional debconf questions in ubiquity to match up against partman-auto/text/{use_device,resize_use_free,custom_partitioning} [17:37] Something that I'll work on tomorrow. [18:02] user-setup: evand * r147 ubuntu/ (debian/changelog user-setup-apply user-setup-ask): [18:02] user-setup: apt-install ecryptfs-utils in user-setup-ask rather than -apply to [18:02] user-setup: better accommodate ubiquity. [18:02] ^ I tested that in both ubiquity and d-i [18:02] Once the new ecryptfs-utils hits the CD and we upload a new ubiquity, that should fix the encryption crash [18:03] great [18:06] user-setup: evand * r148 ubuntu/debian/changelog: releasing version 1.23ubuntu8 [18:09] cjwatson: I filed that bug report, by the way if i put the preseed preseed/late_command string "in-target sed -i ... " I get exit code 1 in the installer [18:11] cjwatson: should it be in-target " [18:14] kirkland: can you look at bug 321345? it has: [18:14] Jan 25 23:18:06 finish-install: cat: can't open '/dev/shm/.ecryptfs-jessie': No such file or directory [18:14] Launchpad bug 321345 in user-setup "Alternate Jaunty Alpha 3 CD doesn't automatically add user to sudoers" [Undecided,Incomplete] https://launchpad.net/bugs/321345 [18:14] Jan 25 23:18:06 finish-install: warning: /usr/lib/finish-install.d/06user-setup returned error code 1 [18:14] which manifests as sudo not being connfigured [18:14] -n [18:15] DogWater: no, it should not be in-target " [18:15] * kirkland looking [18:15] DogWater: do you have any more detail than exit code 1, maybe from the logs? [18:15] let me run through it again [18:16] I suspect we may have reached the point where the bug is in fact in the late_command value :-) [18:16] but we'll have to see [18:18] well, the command works great if i just type it in in the prompt at the end of the install [18:18] so i dunno about that [18:18] remember that the shell does not, by default, report non-zero exit statuses [18:18] so that doesn't actually prove anything ... [18:18] remind me what the command is? [18:19] preseed preseed/late_command string "in-target sed -i 's/http:\/\/10.1.0.1 [18:19] \/ubuntu/http:\/\/mirrors.osuosl.org\/ubuntu/g' /etc/apt/sources.list; in-target [18:19] apt-get update" [18:20] its all one line though [18:20] in the actual document [18:20] cjwatson: i find this curious: The home directory `/home/jessie' already exists. Not copying from `/etc/skel' [18:21] DogWater: might there be any logs indicating that apt-get failed to download some Packages files? [18:22] DogWater: if you want to make it succeed even if that happens, make it "...; in-target apt-get update || true" [18:24] cjwatson: yes it said that it couldn't download some source stuff [18:25] DogWater: right, you probably made some kind of mistake in your mirror then, but you can use || true to make the install proceed anyway [18:29] cjwatson: just so im clear i also need the sources as well? [18:29] cjwatson: this is currently what I have: main,main/debian-installer,restricted,restricted/debian-installer,universe,multiverse [18:30] kirkland: it is odd, but the target filesystem appears to have been clean beforehand, I think ... [18:30] DogWater: source vs. binary is independent of sections [18:30] cjwatson: so it would be $dist-source? [18:30] cjwatson: i'm still digging ... [18:32] DogWater: 'man debmirror' and search for "source" says that it should include source by default. (And no, it isn't $dist-source. Source indexes land in dists/$DIST/$COMPONENT/source/ where $DIST might be jaunty and $COMPONENT might be main; the actual source packages would end up in pool/ with everything else) [18:33] DogWater: though surely the question here is whether mirrors.osuosl.org/ubuntu has a working source mirror [18:33] ? [18:33] according to debmirror it downloaded everything that it intended to download [18:33] i used the easynews mirror because it was faster [18:34] but the apt-get update here is running *after* you've substituted /etc/apt/sources.list to point to mirrors.osuosl.org [18:34] so is your own mirror actually involved? at any rate you need to investigate this at your end since you're armed with your log files and access to your mirror, and I'm not :) [18:35] yeah, i get what you're saying, sorry i'm really frazzled today [18:36] so perhaps just running apt-get update and then changing the lines would be more appropriate [18:36] I don't see why [18:36] because my mirror is complete [18:36] the point of apt-get update is to cope with you having changed the lines [18:36] ah [18:36] if your mirror is complete, why are you running sed over /etc/apt/sources.list at all? [18:36] because the 10.x network isn't accessible from the production network [18:37] therefore you have to sed sources.list and then run apt-get update; no way around that [18:37] but you should look at the resulting sources.list and the log files, and make sure that your sources.list is in fact correct ... [18:37] right, it doesn't actually change at this point [18:39] cjwatson: yeah, alrighty i'll try and see what the deal is, for some reason the sed isn't firing but the apt-get update is firing and its trying to download files from the mirror that arent there but im guessing should be there ;-) [18:40] so i'll figure all that out, thanks [18:40] kirkland: it's not entirely impossible that the target home directory already exists. Maybe we should refuse encrypted-home if that's the case? [18:42] cjwatson: i'll test that; i'm downloading today's alternate iso now [18:44] cjwatson: i'm looking at the adduser code [18:44] cjwatson: doesn't look like to me that anything gets short-circuited that would keep the ecryptfs-setup-private to not run [18:45] cjwatson: can you confirm that adduser is emitting the error message, "The home directory `/home/jessie' already exists. Not copying from `/etc/skel'." ? [18:45] cjwatson: i think i've found the line that matches that [18:45] yes [18:45] and that indeed causes ecryptfs-setup-private to not run [18:46] cjwatson: oh, i see [18:46] cjwatson: yeah, i only execute in the last else{} block [18:47] it's tricky for an existing home directory isn't it? [18:47] cjwatson: yeah totally [18:47] cjwatson: b/c we'd mount on top of that [18:47] cjwatson: and i can't very well do a migration of data at that point [18:47] cjwatson: aufs/overlay magic might help [18:47] cjwatson: but that's very non-trivial :-) [18:47] so user-setup will need to say "sorry Dave, I can't do that" and then carry on unencrypted, IMO [18:48] cjwatson: that's fine by me [18:49] cjwatson: it could suggest that the user could have an encrypted-private setup, after the install [18:49] cjwatson: ie, i could handle that in the wiki docs [18:49] cjwatson: or in your error message, whatever [18:51] cjwatson: sed: -e expression #1, char 13: unknown option to `s' character 13 is a '.' [18:51] and i'm not entirely sure why it would say -e when its called by -i, but i'll keep looking at it [18:52] cjwatson: k, i added comments to that effect in the bug [18:55] DogWater: -e is implicit [18:57] ah, kickseed is eating the escaping :-( [18:57] bloody thing [18:57] preseed preseed/late_command string "in-target sed -i 's,http://10.1.0.1/ubuntu,http://mirrors.osuosl.org/ubuntu,g' /etc/apt/sources.list; in-target apt-get update" [18:57] avoiding the repeated \/ is probably a good idea anyway! [19:00] so wait you dont have to escape the '.'? [19:01] strictly yes, but it's only a problem if you're likely to have something else matching the regular expression http://10.1.0.1/ubuntu ... [19:01] oh, because it was complaining about the ',' [19:01] err . [19:01] the regular expression metacharacter "." matches the single character "." [19:01] that was a red herring [19:01] but i'll try it the way you pasted above [19:02] it was because the // after http: got accidentally unescaped, and so it thought it was a sed command like this: [19:02] s/http://10.1.0.1... [19:02] so it tried to interpret "10.1.0.1..." as s/// options, and gave up at the first . [19:03] since a number after s/// is valid (it means replace only the th match) [19:05] man the amd64 installer blew up in my face a bit ago, im gonna try it on a different machine [20:28] cjwatson: im assuming if i get an error in the install saying that there are no kernel modules available when i go to install the amd64 version of intrepid that my mirror is hosed? [20:30] it means that the installer initrd is out of sync with the archive [20:31] perhaps because you're using an initrd from -updates and your mirror only has the original release [20:31] ah i see im using your new initrd [20:33] has that one been pushed out to anywhere yet? [20:34] i just tried it with anl.gov's mirror which is supposed to be the most up to date one and it doesn't have it either [20:34] yes, to -propossed [20:34] -proposed [20:34] archive.ubuntu.com is the most up to date one [20:34] I posted the URLs in the bug; if mirrors aren't picking them up I'm afraid I have no way of diagnosing or helping [20:37] yeah, archive.ubuntu.com doesn't have it either [20:46] cjwatson: is there some way I can make my mirror work with it? [22:36] hi [22:37] I have a problem with synaptic. Can I ask for help?