UbuntuIRC / 2008 /06 /24 /#ubuntu-installer.txt
niansa
Initial commit
4aa5fce
=== cjwatson_ is now known as cjwatson
[02:35] <CIA-48> console-setup: cjwatson * r71 ubuntu/debian/ (17 files in 2 dirs): merge from Debian 1.25
[02:37] <CIA-48> console-setup: cjwatson * r72 ubuntu/debian/changelog: releasing version 1.25ubuntu1
[08:11] <georg> Hi cjwatson
[08:11] <georg> this is georg
[08:12] <georg> i am trying to install the desktop cd it's giving me the error like
[08:12] <georg> Exception in GTK frontend (invoking crash handler):
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: Traceback (most recent call last):
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: File "/usr/lib/ubiquity/ubiquity/frontend/gtkui.py", line 1494, in watch_debconf_fd_helper
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: return callback(source, debconf_condition)
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: File "/usr/lib/ubiquity/ubiquity/filteredcommand.py", line 177, in process_input
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: if not self.process_line():
[08:12] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: File "/usr/lib/ubiquity/ubiquity/filteredcommand.py", line 110, in process_line
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: return self.dbfilter.process_line()
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: File "/usr/lib/ubiquity/ubiquity/debconffilter.py", line 231, in process_line
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: if not input_widgets[0].run(priority, question):
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: File "/usr/lib/ubiquity/ubiquity/components/summary.py", line 55, in run
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: except DebconfError:
[08:13] <georg> Jun 24 06:46:09 ubuntu ubiquity[7244]: NameError: global name 'DebconfError' is not defined
[08:13] <georg> can u guide me how to rectify this
[08:14] <georg> cjwatson: can u guide me how to rectify this
[08:19] <georg> cjwatson: can u guide me how to rectify this
[08:43] <georg> cjwatson: can u able to reply me for this problem
[09:45] <xivulon> morning davmor2
[09:45] <xivulon> was home late yesterday and while I managed to reproduce couldn't do much debugging
[10:05] <davmor2> xivulon: it's interesting though that the iso worked flawlessly it was only the cd that failed
[10:24] <georg> Hi Cjwatson
[10:50] <cjwatson> georg: please don't repeat your problem over and over again
[10:50] <cjwatson> georg: your problem is fixed in Ubuntu 8.04
[10:51] <cjwatson> georg: it's bug 177442
[10:51] <georg> cjwatson: sorry for that
[10:52] <georg> foe sending the problem again and agian
[10:53] <georg> Thnaks for sending the reply i will go through the link
[10:57] <xivulon> davmor2 that is how I suspected, since I normally always tested the ISO and it used to worked
[10:59] <xivulon> I am quite sure that it is the same issue as we had in partman-auto-loop, I.E. some process grepping /proc/mounts by device AFTER grub-installer
[10:59] <xivulon> cjwatson could that be ^?
[11:00] <cjwatson> I'm sorry, I have no context for this
[11:01] <xivulon> cjwatson we are investigating a ubiquity error when wubi is installed from CD (i.e. when it boots off files extracted into /ubuntu/install/cdrom)
[11:01] <xivulon> what happens is that in this case /isodevice/ubuntu/install/cdrom is bindmounted to /cdrom in the initrd (casper)
[11:02] <CIA-48> yaboot-installer: cjwatson * r261 ubuntu/debian/ (changelog control): Set Vcs-Bzr for Ubuntu.
[11:02] <xivulon> hence if you grep /proc/mounts filtering for the root device you get two lines instead of one
[11:03] <xivulon> davmor2 would you mind providing a link again to installation-logs.zip?
[11:04] <davmor2> xivulon: should bw www.davmor2.co.uk/installation-logs.zip if memory serves
[11:04] <davmor2> indeed it is :)
[11:04] <xivulon> thx
[11:04] <davmor2> AFK
[11:11] <cjwatson> xivulon: certainly seems plausible
=== davmor2 is now known as davmor2_away
[11:38] <CIA-48> partman-basicfilesystems: cjwatson * r567 ubuntu/ (62 files in 3 dirs): merge from Debian 60
[11:41] <CIA-48> partman-basicfilesystems: cjwatson * r568 ubuntu/debian/changelog: releasing version 60ubuntu1
[12:04] <CIA-48> net-retriever: cjwatson * r354 ubuntu/ (6 files in 3 dirs): merge from Debian 1.22
[12:57] <xivulon> can other packages call ubiquity/scripts/install.py > statusChange(...) ?
[12:58] <cjwatson> that's not a public interface, no
[12:58] <cjwatson> things called by install.py should use the normal debconf progress interface
[12:59] <xivulon> strange it seems that the only call comes from updateInterface, but that uses float(percent)
[13:00] <cjwatson> what are you trying to do?
[13:00] <xivulon> so how can the casting int(percent) within statusChange fail
[13:00] <xivulon> tracking the error by reading the code
[13:00] <xivulon> in my lunch break
[13:01] <cjwatson> would need to know what percent was, I guess
[13:01] <cjwatson> there are probably some floats that aren't castable to int, like not-a-number type things
[13:01] <cjwatson> or infinity
[13:01] <cjwatson> exactly how you'd get those from a string I'm not quite sure
[13:01] <xivulon> ah of course
[13:02] <cjwatson> >>> int(float('infinity'))
[13:02] <cjwatson> Traceback (most recent call last):
[13:02] <cjwatson> File "<stdin>", line 1, in <module>
[13:02] <cjwatson> OverflowError: cannot convert float infinity to long
[13:02] <cjwatson> still, I'd be extremely surprised if apt sent back infinity
[13:04] <xivulon> also because the error should probably read "cannot convert float infinity to long"
[13:06] <cjwatson> what error are you getting?
[13:07] <xivulon> http://paste.ubuntu.com/22590/
[13:07] <cjwatson> oh, you misunderstood the trace
[13:07] <cjwatson> the error is not from casting to int
[13:08] <cjwatson> I think you're reading the trace upside-down - the innermost frame is at the bottom
[13:09] <cjwatson> not sure what requires an integer there though
[13:20] <xivulon> ops yes
[13:26] <xivulon> well even worse, ("%s %s\n" % (command, ' '.join(map(str, params)))) what has that to do with ints?
[13:26] <xivulon> hmm lunchtime almosto over
=== davmor2_away is now known as davmor2
[13:55] <davmor2> xivulon: is there an new wubi to test yet?
[13:56] <xivulon> davmor2: not really, hadn't have time to debug yesterday much, and my little trial during lunch didn't go to far either
[13:57] <xivulon> would need more debugging info, hopefully evand made some progress
[13:57] <xivulon> davmor2: it would still help a lot if you could check the other bugs with an ISO installation
[13:58] <xivulon> and provide feedback in each of the bugs
[13:58] <davmor2> xivulon: it just worked :)
[13:58] <xivulon> davmor2: here is a list of the relevant bugs
[13:58] <xivulon> http://ubuntuforums.org/showpost.php?p=5230609&postcount=2
[13:58] <davmor2> But I'll do a quick tag through the bugs and see if any are still about :)
[13:59] <xivulon> in particular LP217348
[14:00] <xivulon> it is sufficient to install on a secondary disk (USB might do)
[14:00] <davmor2> xivulon: it did boot I have multiple hds 2 are on usb but they are detected and mounted
[14:01] <xivulon> davmor2: it is important to check whether wubi works when it is installed on (hd1) in grub terms
[14:01] <davmor2> xivulon: right so wubi needs to install on the second drive?
[14:02] <xivulon> yes please and then post a comment on 217348 as steve is waiting for it
[14:02] <davmor2> NP's
[14:03] <xivulon> much appreciated
[14:05] <xivulon> To test 204128 wubi 503 has to be on a CD or you need to create your own CD by unpacking, replacing wubi.exe and packing back the ISO, then burning a CD
[14:05] <xivulon> that requires vista with latest patches
[14:05] <xivulon> and wubi-selfextract.exe
[14:12] <davmor2> xivulon: is there a selfextract link?
[14:15] <xivulon> davmor2 http://wubi-installer.org/devel/minefield/wubi-selfextract.exe
[14:15] <xivulon> I had posted the instructions for repackaging an ISO somewhere...
[14:16] <davmor2> xivulon: so does the self extract need to be on the cd or the 503? I'm losing the tread a bit?
[14:16] <xivulon> self-extract is rev503 but repackaged to that it is compressed self-extracting
[14:17] <davmor2> xivulon: I just use iso master it's a gui for editing iso's :)
[14:18] <xivulon> that will do
[14:18] <xivulon> ah the manual instructions are in the same bug: https://bugs.edge.launchpad.net/wubi/+bug/204128/comments/19
[14:19] <davmor2> no probs
[14:23] <xivulon> davmor2, cd extraction re 207137 / 230716 is ok (apart of the fact that we hit the wall after grub-installer...) correct?
[14:24] <xivulon> would you mind commenting in those bugs?
[14:26] <xivulon> davmor2, testing 236741 is also easy, when you install on a secondary drive, create a /ubuntu directory on the primary drive with some dummy files, uninstall and check that the first dir does not get deleted
[14:27] <xivulon> for 223155 you need to run wubi (windows) with less than 256MB, you should see a warning but be able to continue the installation
[14:28] <xivulon> I guess that is about it
[14:28] <davmor2> xivulon: that could be an issue lowest spec machine I have has a gig
[14:28] <xivulon> np
[14:29] <xivulon> evand can probably check that easily in vm
[14:32] <evand> indeed, added to my todo list.
[14:32] <xivulon> hi evand, any luch with the ubiquity issue?
[14:33] <evand> not yet, though I have a few ideas that I'm going to try shortly.
[14:34] <xivulon> I didn't go too far myself, only insight is that it only happens when booting off extracted CD (/ubuntu/install/cdrom) while it works well when booting off an ISO
[14:35] <xivulon> I cannot run any test here, but if you can narrow down the scope I can try to go through the code and see if anything jumps out
[14:37] <evand> which I find to be horribly odd. Perhaps this is a red herring.
[14:40] <xivulon> evand as explained above, I think some process might try to grep /proc/mounts by device, and it ends up with 2 lines instead of 1
[14:41] <xivulon> I would not think the setup makes any difference other than for the bindmount
[14:43] <xivulon> you might want to try to use a symlink in caper check_device as opposed to bindmount
[14:48] <davmor2> xivulon: cd re-created and burning I'll hit the nfs unclean with vista too:)
[14:54] <xivulon> evand, see if this helps: http://paste.ubuntu.com/22617/
[14:55] <xivulon> mostly a blind shot but it might help narrow down the issue
[14:58] <davmor2> xivulon: out of curiosity the 236741 but would that work on a second partition too?
[14:59] <xivulon> davmor2: yes, to test it you need to have two \ubuntu directories, on any "drive letter"
[15:00] <xivulon> the uninstaller obviously only has to remove the correct one
[15:00] <davmor2> xivulon: cool I can test that on the vista machine too then :) Ah one cd so many uses :)
[15:00] <xivulon> as usual please add a comment to the bug report
[15:01] <davmor2> just checking vista for updates first :)
[15:01] <davmor2> np's
[15:01] <xivulon> that's like godot
[15:01] <CIA-48> partman-base: cjwatson * r95 ubuntu/ (11 files in 5 dirs): merge from Debian 121
[15:22] <davmor2> xivulon: vista up-to-date and partitioned now for wubi
[15:22] <xivulon> cool
[15:25] <davmor2> grrr wubi-selfextractor couldn't find wubi.exe
[15:25] <xivulon> ?
[15:26] <davmor2> xivulon: Windows cannot find 'wubi.wxw'. Make sure you typed the name correctly, and then try again
[15:26] <davmor2> s/wxw/exe
[15:27] <davmor2> does the extractor need wubi.exe on the cd too?
[15:29] <xivulon> hm no the wubi-self extract is simply wubi compressed with 7z in self extracting mode. i.e. you should be able to run wubi-selfextract from hard disk as well just to test
[15:30] <davmor2> xivulon: ah hang on does it need to be called wubi-selfextract or can it be called wubi.exe?
[15:31] <xivulon> hmm I wouldn't think it matters, but I might have changed the makefile in 503 so that the internal name is different...
[15:32] <xivulon> let me check
[15:32] <davmor2> I called it wubi.exe so it fell in line with umenu and then remembered that umenu wasn't working#
[15:33] <xivulon> ah yes that is my fault
[15:34] <xivulon> skip that test for now, I will produce anothe self extract later on today
[15:34] <davmor2> when can we expect wubi-selfextract 504 then :)
[15:34] <xivulon> tonight
[15:34] <xivulon> or earlier with evand help
[15:35] <davmor2> mail me at davmor2@gmail.com when It's up and get it asap and test it then :) in the mean time I'll drop 503 on a cd a test with that instead okay
[15:44] <xivulon> evand can you please apply the patch in https://bugs.edge.launchpad.net/wubi/+bug/242682 and rebuild wubi-selfextract?
[15:46] <CIA-48> anna: cjwatson * r411 ubuntu/debian/ (changelog po/am.po po/ml.po po/mr.po po/pa.po): merge from Debian 1.32
[15:47] <CIA-48> anna: cjwatson * r412 ubuntu/debian/changelog: releasing version 1.32ubuntu1
[15:48] <xivulon> davmor2 in the meantime can you please test LP217348 LP207137+230716 236741
[15:49] <davmor2> yeap on it I'll still work through as much of the list as poss :)
[15:49] <davmor2> just made a new cd with 503 on it
[15:50] <xivulon> a CD won't help without a working selfextract..
[15:51] <davmor2> Test on my xp box I'll try the vista either tonight or tomorrow :)
[15:52] <xivulon> great thanks :)
[15:53] <xivulon> evand we'll need the localized version of XP for a few tests: 136682 222690 223250
[15:54] <evand> 222690> You need someone with the French version of XP Home, which I do not have nor can I acquire, unfortunately.
[15:55] <CIA-48> partman-ext3: cjwatson * r737 ubuntu/debian/ (po/mr.po changelog po/ml.po po/pa.po): merge from Debian 52
[15:57] <CIA-48> partman-ext3: cjwatson * r738 ubuntu/debian/changelog: releasing version 52ubuntu1
[15:57] <evand> xivulon: davmor2: http://people.ubuntu.com/~evand/wubi - built and uploaded
[15:58] <davmor2> evand: sorry is this the wubi-selfextract?
[15:58] <evand> Yes
[15:58] <davmor2> evand: cool ta
[15:59] <evand> You're welcome
[16:09] <superm1> evand, do you think you will still be able to get around to adding the fix for bug 219192 in?
[16:10] <xivulon> evand thanks a lot
[16:10] <xivulon> evand for 222690 I have already asked verrotin, but no feedback yet
[16:13] <evand> argh, that completely fell off my todo list. No, I think it's too late to fix that in .1.
[16:13] <superm1> oh that's too bad :(
[16:13] <xivulon> evand did you try the casper fix ^ by any chance?
[16:13] <evand> really sorry about that. I'll make sure it gets in .2 as soon as it opens.
[16:14] <evand> xivulon: yes and it seems to work, but testing again.
[16:15] <xivulon> evand then it is certainly the bindmount creating issues, so you can either grep for /proc/mounts to find the offending program
[16:15] <superm1> evand, could you make sure that it's at least fixed in intrepid then in the short term so that when the dailies start coming out, they're ok?
[16:15] <xivulon> or leave the fix as is not sure if using symlink in casper as opposed to bindmount has any other sideeffect (cjwatson?)
[16:16] <evand> superm1: indeed, will do
[16:16] <superm1> thanks
[16:17] <cjwatson> symlink from where to where?
[16:17] <cjwatson> I'm sorry, I'm not getting time to follow all of this as it goes past
[16:17] <xivulon> cjwatson http://paste.ubuntu.com/22617/
[16:18] <xivulon> was a quick hack to see if the bindmount theory was correct
[16:18] <evand> http://evalicious.com/tmp/wubi-20080623-syslog-2.txt <- where the bug manifested.
[16:18] <xivulon> but if the bindmount creates all those issues we might as well use a symlink unless there are objections of course
[16:20] <cjwatson> you need to be careful that the symlink will actually be resolvable in the new root (symlinks are resolved when they're used, not when they're created); furthermore, other things may well expect to be able to unmount /cdrom later, so you need to do end-to-end tests before doing this
[16:21] <xivulon> those were the sort of objections I was looking for :P
[16:22] <davmor2> xivulon: evand: wubi is still dropping into initramfs on a dirty xp drive
[16:22] <davmor2> s/xp/ntfs
[16:22] <xivulon> davmor2: yes that is expected
[16:22] <cjwatson> bind-mounts are resolved at mount time, so they avoid the typical sorts of issues you get when creating symlinks from the initramfs (which by definition is in a different root filesystem)
[16:22] <xivulon> as we don't have fsck.ntfs
[16:23] <xivulon> davmor2: you should at least see a more meaningful message, at least in recovery mode
[16:24] <davmor2> it's not in normal mode
[16:24] <davmor2> recovery yes
[16:26] <xivulon> cjwatson: well at least now we know that the bindmount is the issue, so grepping for /proc/mounts should help narrow down the options
[16:31] <xivulon> hmm don't see how that could work with a symlink: ./scripts/casper-bottom/05mountpoints:mount -n -o move /cdrom /root/cdrom
[16:32] <xivulon> evand are you sure it works?
[16:33] <evand> I'm in the process of testing it a second time.
[16:33] <davmor2> xivulon: 217348 this seems to be resolved now isn't it? this is why you wanted a copy of the menu.lst isn't it?
[16:34] <xivulon> should be
[16:35] <davmor2> you were happy with the menu.lst yesterday so I'm going to assume it's fixed looks the same today :)
[16:35] <xivulon> davmor2 can you please add some more context about the test, explain for instance that you installed wubi on a secondary hard disk and still booted (provide relevant menu.lst)
[16:36] <xivulon> the test to be valid should be about a wubi installation on hard disk 2 as you might end up booting off hard disk 1 anyway since that is a normal fallback
[16:37] <davmor2> Argh mis-read it sorry I'll add in a bit need to do that one on the vista machine. My own fault.
[16:38] <davmor2> 224311 I don't have issues with
[16:39] <davmor2> 227023 doesn't effect me either
=== xivulon_ is now known as xivulon
[16:41] <xivulon> davmor2: 224311 cannot be reproduced unless you have the buggy bios... but it important to check that the new grub4dos does not create regressions
[16:41] <davmor2> np's here :)
[16:41] <xivulon> same story for 227023
[16:43] <davmor2> xivulon: other tests will need to wait till latter or in the morning depending on when I get back
[16:43] <xivulon> does selfextract work now?
[16:43] <xivulon> np
[16:44] <xivulon> thanks a bunch for everything
[16:44] <xivulon> ops too late
[16:45] <evand> yeah, you're right. Symlinks don't work as apt-setup wants to unmount and remount it.
[16:46] <xivulon> plus the mount -o move...
[16:46] <xivulon> so we know as much as before :(
[16:47] <xivulon> on a different topic, isn't an initrd panic message supposed to be visible also when not in recovery mode?
[16:47] <evand> well no, we know that it does work (the install can proceed beyond that point without issue), so we know it's definitely an issue with using a bindmount for some strange reason.
[16:47] <xivulon> I can see that usplash gets killed
[16:48] <evand> Not sure offhand.
[16:49] <xivulon> hmm but doesn't casper bottom 05 try to mount move /cdrom to /root/cdrom? surely that must fail when cdrom is a symlink
[22:32] <xivulon> evand any progress?
[22:51] <xivulon> evand I updated to wubi 504 to fix 242682
[22:52] <bdmurray> evand: is there a test case for bug 241420 at all?
[22:52] <bdmurray> xivulon: hi!
[22:52] <xivulon> bdmurray, hi
[22:53] <xivulon> bdmurray, I commented on 204128 testing
[22:54] <xivulon> as you can see that would require creating an ISO (must be latest daily 8.04.1) with the new wubi-selfextract
[22:54] <bdmurray> right, I'd forgotten about using wubi.exe on the hard drive not being a good test. :(
[22:55] <xivulon> well np, wubi-selfextract 503 was not working anyway :P
[22:55] <xivulon> new one (504) should be good
[22:56] <xivulon> note that the linux side installation will not complete, but that should not be relevant for that bug
[22:56] <xivulon> linux side installation now only works when using an iso