/srv/irclogs.ubuntu.com/2009/02/27/#ubuntu-installer.txt

CIA-3localechooser: cjwatson * r146 ubuntu/debian/changelog: releasing version 2.09ubuntu204:35
cjwatsonstgraber: thanks, but he added set -x to the wrong script. See my comment04:42
shtylmanooo... that evil thing that takes 10000 million years to generate templates!04:42
CIA-3partman-partitioning: cjwatson * r698 ubuntu/debian/changelog: releasing version 64ubuntu404:48
CIA-3partman-auto-lvm: cjwatson * r217 ubuntu/debian/changelog: releasing version 32ubuntu404:49
CIA-3tasksel: cjwatson * r1397 ubuntu/debian/changelog: correct changelog05:15
CIA-3tasksel: cjwatson * r1398 ubuntu/debian/changelog: releasing version 2.73ubuntu1505:16
CIA-3user-setup: cjwatson * r160 ubuntu/ (3 files in 2 dirs):05:27
CIA-3user-setup: user-setup/password-weak can't be preseeded since that would interfere05:27
CIA-3user-setup: with correct interactive behaviour, but create05:27
CIA-3user-setup: user-setup/allow-password-weak which can.05:27
CIA-3user-setup: cjwatson * r161 ubuntu/debian/user-setup-udeb.templates: move template to live with the other preseedable-only templates05:29
CIA-3user-setup: cjwatson * r162 ubuntu/ (4 files in 2 dirs):05:32
CIA-3user-setup: Rename passwd/allow-password-empty to user-setup/allow-password-empty,05:32
CIA-3user-setup: for consistency with user-setup/password-empty. Provide compatibility05:32
CIA-3user-setup: with the old name for one release.05:32
CIA-3ubiquity: cjwatson * r3067 ubiquity/ (debian/changelog ubiquity/frontend/base.py):05:36
CIA-3ubiquity: Cope with renaming of passwd/allow-password-empty to05:36
CIA-3ubiquity: user-setup/allow-password-empty in user-setup 1.23ubuntu14.05:36
CIA-3user-setup: cjwatson * r163 ubuntu/debian/changelog: releasing version 1.23ubuntu1405:38
TheMusosomeone's up early. :)05:40
cjwatsonyeah, have a cough and couldn't sleep05:44
StevenKcjwatson: I hate it when that happens. :-(05:47
TheMusoAh that sucks.05:48
CIA-3ubiquity: superm1 * r3068 ubiquity/ (debian/changelog ubiquity/components/mythbuntu_install.py): Remove dead AdditionalDrivers code06:48
CIA-3debian-installer-utils: cjwatson * r668 ubuntu/ (14 files in 3 dirs): merge from Debian 1.6606:50
CIA-3debian-installer-utils: cjwatson * r669 ubuntu/debian/changelog: releasing version 1.66ubuntu106:52
tjaaltoncjwatson: trying the iso now, but I'd like to use our preseed-file as well. fetching it fails, because the installer messes up the url somehow07:51
tjaaltonit adds "BOOT_IMAGE=/install/vmlinuz" to the url :)07:51
tjaaltoncjwatson: but is it the new udev that should fix the issues with partman?08:05
cjwatsonI think it ought to, yes. I don't think I changed the installer in any other way08:19
tjaaltonbut it'd need a rebuild of d-i to generate a netboot image I could use?08:26
cjwatsonI can build you a netboot image08:28
tjaaltonthanks!08:28
markcjwatson: did you see the > 16 disks bug? no idea how to assign it to you, but I subscribed you08:32
cjwatsonI have it now, thanks08:33
cjwatsontjaalton: which netboot files do you normally use?08:34
tjaaltoncjwatson: just the tarball, in this case amd6408:35
tjaaltonnetboot.tar.gz08:35
cjwatsonoh, um, not sure I can easily do amd6408:35
cjwatsongive me a bit, then, I'll see08:35
tjaalton32bit is fine08:36
tjaaltonjust need to edit the pxeboot file08:36
tjaaltonno biggie08:36
cjwatsontjaalton: ok, amd64 turns out to be more painful than I'd like ... http://people.ubuntu.com/~cjwatson/tmp/netboot.tar.gz is an i386 build08:51
cjwatsontjaalton: if you want amd64, build udev with http://git.kernel.org/?p=linux/hotplug/udev.git;a=commitdiff;h=3d3a0a709a38805259fe07240c3ca47a120dd5d6 applied, copy the resulting udev-udeb into d-i's build/localudebs/, and 'fakeroot make rebuild_netboot' in the build/ directory08:51
tjaaltoncjwatson: thanks, trying the image now08:57
tjaaltonhmm, still fails to remove the old logical volumes09:00
davmor2tjaalton: you still having issues?09:01
tjaaltonyes09:01
cjwatsonok, that could be another problem I do know about09:02
cjwatsonthe LVM tools aren't really atomic in several ways09:02
tjaaltonI'm trying to install lvm on a clean disk now09:02
cjwatsonwhen you run 'vgchange -a n ubuntu && vgremove ubuntu', vgchange only tells the kernel to deactivate the logical volumes, but doesn't actually wait until userspace has finished doing so09:03
cjwatsonand, with the current architecture, if you made it try to wait, it would deadlock09:03
tjaaltonoh..09:03
cjwatsonso I plan to stick some udevadm settle calls into the installer to work around this09:03
cjwatsonI've had a couple of long conversations with Scott about this over the last few days, and I understand the problem much better now09:03
davmor2tjaalton: I've come on to late what are trying do install over an lvm install?09:04
cjwatsonhttp://paste.ubuntu.com/123706/ applied to /lib/partman/lib/lvm-base.sh is something like what I plan to do09:05
tjaaltondavmor2: hmm, sorry I don't follow?09:06
tjaaltontrying to install lvm on a clean disk now complains that the volume group name is already in use09:06
* cjwatson hands davmor2 some punctuation :-)09:06
cjwatsonhow clean?09:07
tjaaltonI deleted the partitions before rebooting09:07
cjwatsonI wonder if LVM metadata was left around in the old partitions09:07
cjwatsonI suspect we need to be better about cleaning that out when creating new partitions09:08
tjaaltonI'll try harder :)09:08
tjaaltonbad fdisk.. nothing like on tru6409:09
cjwatsonhmm, so the difficulty is a layering one09:13
cjwatsonwe don't want to clear out LVM metadata unconditionally, only when the PV partitions are newly-created09:13
tjaaltonI tried the iso without any preseeding, and then it succeeded in installing on a clean disk09:18
tjaaltonso..09:18
tjaaltonone note about the succeeded partitioning; it showed the maximum size of the vg, and when I chose it I got an error message saying that there's not enough space09:31
tjaaltonso I had to change the value09:32
cjwatsonthat's fixed09:32
tjaaltonnice09:32
cjwatsonbug 33464809:32
ubottuLaunchpad bug 334648 in partman-auto-lvm "Install fails when choosing a size equals to the max size" [High,Fix released] https://launchpad.net/bugs/33464809:32
tjaaltonah09:32
cjwatsonwas a rounding error, didn't happen to me in initial testing because in my test it rounded down rather than up09:32
evandbug 330979 is perplexing.09:37
ubottuLaunchpad bug 330979 in ubiquity "install stops crc error" [Undecided,Confirmed] https://launchpad.net/bugs/33097909:37
evandcjwatson: do you recall why we decided to filter all disks with mounted partitions from parted_devices rather than just /cdrom?09:46
evandIt seems to be creating a headache for a number of users, and I'm wondering if we should try to unmount the partitions in question, then try again, or filter less out.09:46
CIA-3debian-installer: cjwatson * r995 intrepid-proposed/ (3 files in 2 dirs): Move mainline architectures to 2.6.27-13 kernels.09:48
cjwatsonevand: because if any partition is mounted on a disk then you can't tell the kernel to reread its partition table. We added partman/filter_mounted so that they can control it09:49
evandoh, of course09:49
cjwatsonI think this is a bit of a lose-lose situation09:49
cjwatson330979> astonishingly bad bug triage there09:50
evandUgly, but perhaps a dialog warning them that they wont see the following disks because they have mounted partitions09:51
cjwatson"this bug has some similarity to <completely different bug>"09:51
evandlol09:51
cjwatsonor a question asking them whether they want to unmount those partitions or not; of course they may have documents open on them ...09:52
cjwatsonfeels like perhaps the sort of thing that should be done in ubiquity rather than partman-base09:52
cjwatsonin d-i it's much less likely that partitions will be mounted09:52
evandindeed09:52
cjwatsonno objection to ameliorating the lose-lose on either side of course :-)09:54
cjwatsonoh god must kill management-speak09:54
davmor2hmm confirming a bug without reproducing it genius :)09:56
evandhaha, old job titles die hard?09:58
davmor2cjwatson: out of curiosity why has lvm become such a bummer10:07
cjwatsondavmor2: udev has got faster and so the lvm race conditions become more obvious10:09
cjwatsonor possibly slower. but at any rate perturbed it enough that more people happen to encounter it. :-)10:09
davmor2:) Okay10:10
cjwatsonevand: bug 292159: last week you said you'd work on it "next week"10:10
ubottuLaunchpad bug 292159 in linux "MASTER update-initramfs is disabled since running on a live CD but it is running from a flash drive. " [Undecided,New] https://launchpad.net/bugs/29215910:10
evandargh, indeed.  I'll get on it in a few minutes.  I'm just in the process of putting our previous discussion in a bug report and discussing changes to usb-creator with a contributor.10:11
evandbut I'll make it my top priority today10:11
cjwatsonthanks10:12
CIA-3debian-installer: cjwatson * r996 intrepid-proposed/debian/changelog: releasing version 20080522ubuntu2810:19
davmor2evand: Just to let you know that in ubiquity now everytime you do a guided - whole drive install you get the  m-a page and says it can't find an os to copy from.10:33
evandsome day I will make time to work on that mess of code.10:34
evandbut that bug is one I will fix this release10:34
evandit's part of the jaunty-ubiquity-usability specification10:34
CIA-3usb-creator: evand * r79 trunk/ (debian/changelog gui/usbcreator.glade):10:38
CIA-3usb-creator: "Note that using GTK_WIN_POS_CENTER_ALWAYS is almost always a bad idea."10:38
CIA-3usb-creator: Place the window in the center, don't keep it centered. Thanks Eric10:38
CIA-3usb-creator: Butler.10:38
CIA-3usb-creator: evand * r80 trunk/ (debian/changelog usbcreator/gtk_frontend.py):11:03
CIA-3usb-creator: Set the title property of the labels after creating WrapLabels, not11:03
CIA-3usb-creator: before. Thanks Eric Butler.11:03
CIA-3usb-creator: evand * r81 trunk/debian/ (changelog control): Point Vcs-Bzr in the control file to the proper location.11:43
evandHrm, cjwatson what are your thoughts on having a separate usb-creator icon?  You expressed concern that using the ubiquity icon might lead to confusion, and I'm about to ask Ken if he's willing to create one, should he find some free time, but I wanted to make sure it's something desired by more than just me.12:08
cjwatsonseems sensible for separate programs to have separate icons12:09
evandok, thanks for the input12:09
cjwatsondoesn't seem hugely critical though12:09
evandof course12:11
cjwatsonevand: when you were doing the language_iconview change for oem-config, did you ever figure out why the vertical spacing between cells is so big?12:18
cjwatsonevand: it results in a very tall window, and looks kind of silly on subsequent screens because their contents gets stuck in the middle of this tall window12:18
evandI think my notes for that are long gone.  But I suspect if I did get to the bottom of it, I would have made the change.  Unfortunately as I believe you've pointed out as well, the iconview is quite inflexible.12:19
evandThere are some alternative box packing widgets (either in libview or medsphere), but that seemed to be the cleanest approach.12:20
cjwatsonit seems to just use a cellrenderertext, but I can't quite unpick the twisty mazes of code involved12:30
CIA-3usb-creator: evand * r82 trunk/ (3 files in 3 dirs): Call dd with the correct arguments (LP: #331327). Thanks Martin Pitt!12:38
CIA-3usb-creator: evand * r83 usb-creator/debian/control: Update Vcs-Bzr to the usb-creator-hackers team.12:39
evandugh, I can't reuse the ubuntu-installer list as a contact address for usb-creator in LP12:41
mgariepyhi13:24
stgrabercjwatson: mgariepy is the one doing the preseeds test13:25
mgariepycjwatson : i did try to set -x in /bin/perfrom_recipe but it doesn't log anything13:25
cjwatsonmgariepy: hmm. could you perhaps try putting set -x in /lib/partman/display.d/10initial_auto, /bin/autopartition, and /bin/perform_recipe?13:44
cjwatsonhopefully that should cover it - sorry for having to go round a couple of times13:44
mgariepyno problem13:44
cjwatsonmgariepy: or, actually, why not save time and just put set -x at the top of /lib/partman/lib/base.sh13:44
cjwatsonwill produce honkingly enormous amounts of output13:44
cjwatsonbut the stuff I need should be in there somewhere :-)13:45
mgariepyok i'll put it in base.sh :)13:45
mgariepycjwatson, just uploaded the 15 000 lines syslog :) enjoy13:58
cjwatsonheh13:59
mgariepycjwatson, if you need any more test, just ask  hehe13:59
cjwatsonok, will do; I'm expecting a visitor shortly so may be out for a bit13:59
mgariepyi'll be around for a while14:00
cjwatsonthis looks like it has the necessary bits though, at a first glance14:00
* cjwatson stuffs it into vim rather than a browser14:00
davmor2evand: Re icon just usb stick above a cd :)14:05
cjwatsonmgariepy: aha14:16
cjwatsonmgariepy: you have an invisible trailing space on this line:14:16
cjwatson                      mountpoint{ / }                         \14:16
cjwatsonmgariepy: the backslash has to be the very last character on the line, with no space after it14:16
evanddavmor2: :)14:21
davmor2evand: you know it makes sense :)14:21
mgariepyho14:23
stgrabercjwatson: trying without it now ...14:27
mgariepybut the preseed file was working fine before 8.04.214:29
cjwatsonmgariepy: are you sure it wasn't edited in any way? this is the sort of thing that can easily creep in by accident14:33
cjwatsonI'm pretty certain that none of the preseed-file handling code changed in 8.04.x, and this is right at the bottom level14:34
mgariepyyeah, the preseed file is in subversion ""r1196 | mgariepy | 2008-08-14 18:22:44 -0400 (Thu, 14 Aug 2008) | 1 line""14:36
cjwatsonwell, I'm not sure what could have caused it to work, then, but I know it's broken as-is :-)14:36
mgariepyplus i had like 35 server isntall with it :)14:36
cjwatsonthat's definitely the thing causing the failure - you can see it stopping there in the trace14:37
mgariepybut it's working now, w'll just have to be more carefull when editing preseed. :)14:37
mgariepythanks14:37
mgariepycjwatson, what does we do with the bug ?14:39
stgrabershould we consider that a regression from previous d-i where it used to parse the file and just ignore trailing spaces or was that a parser bug that it used to work ?14:43
cr3_I have the lvm options in a preseed partman-lvm/device_remove_lvm, device_remove_md and confirm. yet, my installation stalled on a bluescreen with this in the logs: Logical volume "LogVol00" successfully removed, Can't remove open logical volume "LogVol01", 0 logical volume(s) in volume group "VolGroup00" now active15:11
cr3_should I report a bug about this or might there be a problem with my preseed?15:11
davmor2cr3_: I hate to say this incase I'm wrong but there is a races issue with lvm cjwatson can fill you in more :)15:12
cr3_davmor2: if it's a known issue, I won't generate more noise by reporting another bug about it. I'll wait for cjwatson's view on the issue15:20
fadercjwatson: ping15:24
fadercjwatson: I'm getting an error when running some server installs about the partition being busy.  It seems to clear up if I try it a second time, but this obviously makes running installs through cron problematic.  (Error message to follow)15:27
faderError informing the kernel about modifications to partition /dev/cciss/c0d0p1 -- Device or resource busy.  This means Linux won't know about any changes you made to /dev/cciss/c0d0p1 until you reboot -- so you shouldn't mount it or use it in any way before rebooting.15:27
fadercr3 suggested I ask you if you knew a workaround...15:27
cr3_... or if this was a known issue :)15:28
cjwatsonstgraber: I am finding it really tough to credit the claim that it worked in 8.04/8.04.1 given that the preseed package literally has not changed at all15:31
cjwatsonstgraber: and given that, to the best of my knowledge, your file should *never* have worked as-is15:31
cjwatsonI honestly think the best answer is to fix the file and move on :)15:32
cjwatsoncr3_,fader: both your problems are part of a general set of bugs being investigated, and that indeed were mentioned in the alpha 5 release notes15:32
cr3_cjwatson: ack, thanks for the reminder about the release notes!15:33
cjwatsonfader: for a server installation, I would appreciate it if you could try http://cdimage.ubuntu.com/tmp/20090226/jaunty-server-i386.iso15:33
cjwatsonthat rolls in a udev change from upstream15:33
fadercjwatson: Will do.  Specifically the -i386 image or will -amd64 work as well?15:33
cjwatsonI am aware that this doesn't fix all the LVM-related bugs15:33
cjwatsonfader: it was built by hand, so I only have i386, sorry15:33
cjwatsonanyway, there should be no need to report a bug; we found out about this before alpha 5 but there wasn't really a way to fix this without significant risk of making things worse15:34
fadercjwatson: Okay, no problem... I'll grab i386.  It's not 100% reproducible so I can only really verify its presence if it happens again15:34
cjwatsonyes, it's a (set of) race condition(s)15:34
cjwatsonthere are two fixes in progress:15:35
cjwatson1) make udev remove inotify watches when devices are removed (udev 139-1 includes this)15:35
cjwatson2) change parts of the installer that call the LVM tools to call udevadm settle in between15:35
faderCool.  I won't file any bugs and I can live with rerunning failed tests by hand for a while.  Thanks, cjwatson :)15:36
CIA-3debian-installer: cjwatson * r1055 ubuntu/debian/changelog:15:39
CIA-3debian-installer: No-change rebuild to pick up new components (particularly udev with the15:39
CIA-3debian-installer: inotify-watch-on-device-remove fix, and debian-installer-utils with15:39
CIA-3debian-installer: update-dev --settle).15:39
mgariepycjwatson, i'll have to make another test monday, the install i was using wasn't using partitionning. :15:41
cjwatsonstgraber,mgariepy: in other words, it is my belief that the file *should* never have worked, even if it accidentally did, which makes it difficult for me to consider it as a regression15:41
mgariepyd-i partman-auto/disk string /dev/sda15:41
mgariepyd-i partman-auto/method string regular15:41
mgariepyd-i partman-auto/choose_recipe select atomic15:41
mgariepyd-i partman/confirm_write_new_label boolean true15:42
mgariepyd-i partman/choose_partition select finish15:42
mgariepyd-i partman/confirm boolean true15:42
mgariepythat's the partitioning i was originaly using.15:42
mgariepyand i assure you that it was working, i'll have more info moday15:42
cjwatson*that* will work15:42
cjwatsonno continuation lines in there15:42
mgariepymonday15:42
cjwatsonit's exclusively the continuation line stuff that broke, AFAICS15:43
cjwatsonthe semantics are the same as in C15:43
cjwatsonwhere you have the same property - backslash-space-newline is different from backslash-newline15:44
mgariepyi'll test and post on the bug monday for more information.15:44
mgariepyif it's not working, should i set -x in base.sh too ?15:44
cjwatsonyeah15:44
mgariepyok :)15:44
cjwatsonprobably the quickest way15:44
mgariepythe quickest way with like 15000 lines hehe15:44
cjwatsonyeah, but consider the round-trip time :)15:45
mgariepyyeah15:45
cjwatsonit's vast overkill for me, certainly - I have to wade through for the few bits I need - but it's good to be pretty much guaranteed success15:45
mgariepywell i'm leaving irc now :) cya15:46
cjwatsonok, thanks15:46
mgariepyhave a nice weekend15:46
CIA-3debian-installer: cjwatson * r1056 ubuntu/debian/changelog: releasing version 20081029ubuntu2215:46
=== cr3_ is now known as cr3
=== cr3_ is now known as cr3
CIA-3usb-creator: evand * r84 usb-creator/ (Makefile debian/changelog):17:15
CIA-3usb-creator: Move the addition of usb-creator.desktop.in to POTFILES.in to the build17:15
CIA-3usb-creator: script.17:15
evandhaving an interesting time trying to get a new initrd to not cause a kernel panic, but will hopefully have a solution soon17:44
MountainXhi-what is the best channel for help installing ubuntu?18:31
MountainXafter I install Ubuntu (8.10, alternate CD 64 bit) on a computer with 5 HDD's, and reboot at the end, the computer fails to boot (black screen, blinking cursor, no message). I think it is looking in the wrong place for the /boot partition. What is the correct way to install a boot partition that isn't on (hd0,0)?18:31
=== fader is now known as fader|lunch
=== fader|lunch is now known as fader
=== saispo_ is now known as saispo

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!