[03:58] <jayvee> Anyone here that has a vague idea of how Wubi works? I’m “remastering” a 9.10 live CD, which works beautifully with Ubiquity. But when I launch Wubi from my remastered disc, the “Install inside Windows” button is gone from the splash screen, whereas it shows up on the 9.10 “gold” image.
[03:58] <jayvee>  So there is some logic in Wubi that makes it check for whether to enable the “Install inside Windows” button or not. I’d like help trying to find out where and what that logic is. :)
[03:59] <StevenK> I *think* it's based on the contents of .disk/info
[04:00] <jayvee> I haven’t touched that at all. I have updated md5sums.txt, because when I did this last year, Wubi kept re-downloading the original image. Found it was because filesystem.squashfs’s md5sum didn’t match. :)
[04:01] <jayvee> Well my .disk folders are the same. .disk/info says “Ubuntu 9.10 "Karmic Koala" - Release i386 (20091028.5)”
[04:01] <jayvee> s/folders/directories/
[04:01] <StevenK> Well, I could be wrong
[04:03] <jayvee> doing a diff -ur between the two structures
[04:03] <jayvee> this caught my attention:
[04:03] <jayvee> File iso-orig/ubuntu is a directory while file iso/ubuntu is a regular empty file
[04:03] <jayvee> ‘ubuntu’ is a symlink that points to -> .
[04:04] <jayvee> I wonder...
[04:12] <jayvee> Ah, just found the Wubi log file in %TEMP%. Unfortunately the entries are datestamped so I can’t do a diff on them.
[04:14] <persia> filter them with awk or cut into new temp files and diff those?
[04:15] <jayvee> Just opened them in gedit and flipped between them really fast. :)
[04:15] <jayvee> Definitely no difference.
[04:15] <jayvee> It seems Wubi is written in Python, and it's bundled within a py2exe launcher.
[04:17] <jayvee> the logfile looks pretty much similar to this: http://launchpadlibrarian.net/31054981/wubi-9.10-rev147.log
[04:17] <persia> Yep.
[04:18] <jayvee> aha
[04:18] <jayvee> https://bugs.launchpad.net/wubi/+bug/461566
[04:18] <ubottu> Launchpad bug 461566 in wubi "kne image plus persistence means that the iso size is too large I believe" [Medium,Confirmed]
[04:19] <jayvee> I run “F:\wubi.exe --force-wubi” and the button magically reappears.
[04:19] <jayvee> aargh
[04:24] <jayvee> From the bug report: “The check in wubi/frontends/win32/cd_menu_page.py (line 63) is too restrictive, that was in place to discriminate against DVD.”
[04:24] <jayvee> uh, discriminate against DVD?
[04:24] <jayvee> sounds fishy to me.
[04:32] <jayvee> bleh, even though the md5sums match, it wants to download the iso still
[04:32] <jayvee> ...over BitTorrent.
[04:32] <jayvee> my desk has a large head-shaped dent in it right now.
[08:40] <ara> ev, is it expected that in ubiquity 2.1.30 the slideshow has no contents?
[08:40] <ev> ara: yeah, it's a known bug
[08:41] <ara> ev, ok, thanks! is there a bug number about it?
[08:41]  * ev digs
[08:43] <ev> ah, I don't believe there is.  I started investigating it the other day when davmor2 mentioned it.
[08:43] <ev> if you have your logs handy, would you be so kind as to file one for me?
[08:43] <ev> otherwise I can fire up a VM
[08:48] <ara> I will file one, don't worry
[08:49] <ara> It is finishing the installation now, when it's done, I will file the bug with logs
[08:49] <ara> ev ^
[08:49] <ev> lovely, thanks ara
[09:05] <dmarkey> can anyone tell me which part of the installer is in charge of detecting consoles? for example /dev/hvc0
[09:28] <cjwatson> dmarkey: finish-install
[09:29] <ara> ev, bug 532502
[09:29] <ubottu> Launchpad bug 532502 in ubiquity "Slideshow has no contents" [Undecided,New] https://launchpad.net/bugs/532502
[09:35] <ev> updated, thanks a bunch!
[09:56] <dmarkey> cjwatson: are there any plans to enable PAE by default?
[09:58] <jayvee> I thought it already was.
[09:58] <jayvee> I installed a VM the other day, and it came with the generic-pae kernel out-of-the-box. Can't remember if it was Karmic or Lucid.
[09:59] <jayvee> In fact, the VM wouldn't boot because of that for some reason. Had to chroot in and install generic instead.
[10:02] <dmarkey> really, hmm
[10:02] <dmarkey> Was it a pre pentium-pro processor?
[10:03] <jayvee> It was a VirtualBox VM, IIRC.
[10:03] <dmarkey> did you enable PAE on the VM?
[10:05] <cjwatson> dmarkey: it's enabled by default if we can detect that it's present
[10:06] <cjwatson> sometimes detection goes a bit wrong
[10:06] <dmarkey> cjwatson: if i download a netboot vmlinuz/initrd, will the vmlinuz be PAE enabled?
[10:06] <cjwatson> not the installer itself
[10:06] <cjwatson> oh, actually, no, I misspoke above
[10:07] <cjwatson> we only default to PAE if you have more than 3GB of memory, and if we can detect that PAE is available
[10:07] <cjwatson> otherwise, PAE is a slowdown
[10:07] <dmarkey> hmm
[10:07] <jayvee> I read somewhere that it gets enabled anyway because NX doesn't work without it.
[10:10] <dmarkey> cjwatson: you see, 32bit kernels only work on xen with PAE enabled.
[10:10] <dmarkey> so, we would have to create a seperate PAE enabled install vmlinuz/initrd
[10:11] <cjwatson> ugh
[10:12] <cjwatson> jayvee: I implemented the relevant bits of code
[10:12] <cjwatson> my misspeaking is merely a reflection of not having had enough coffee yet today
[10:12] <ev> :)
[10:12] <jayvee> heh
[10:12] <cjwatson> dmarkey: don't suppose we can just say "only supported on 64-bit"?
[10:13] <cjwatson> 32-bit has "only" 28 years left to run anyway ...
[10:14] <dmarkey> :)
[10:15] <dmarkey> 2.6.32 is going to be the kernel in the next release, right?
[10:19] <cjwatson> yeah
[10:19] <dmarkey> also, will partition alignment be improved?
[10:20] <cjwatson> yes, if I have anything to do with it; I'm working on that
[10:20] <dmarkey> great!
[10:20] <cjwatson> next step is pulling in parted 2.2
[10:21] <cjwatson> (which is unfortunately another ABI change, I bet)
[10:21] <dmarkey> so are you going to start the first partition at LBA 128 or something?
[10:21] <cjwatson> 1MB I think
[10:22] <cjwatson> might depend on disk topology
[10:22] <dmarkey> well, in a VM you have littl knowledge of the backend disk topology
[10:22] <cjwatson> parted has defaults
[10:22] <cjwatson> which are conservative, in 2.2
[10:22] <cjwatson>   The ped_device_get_*_alignment() functions now return a sane default
[10:22] <cjwatson>   value instead of NULL when the so called topology information is incomplete.
[10:23] <cjwatson>   The default minimum alignment aligns to physical sector size, the default
[10:23] <cjwatson>   optimal alignment is 1MiB, which is what vista and windows 7 do.
[10:23] <cjwatson> and we'll make partman apply optimal alignment constraints to things
[10:24] <dmarkey> sounds good
[10:24] <dmarkey> have you read that VMWare paper on partition alignment?
[10:24] <cjwatson> yes, I believe so
[10:25] <cjwatson> I'm more immediately worried about things like disks that behave grossly suboptimally if the first partition starts at LBA 63, and compatibility with Windows 7; but I think making those work should sort it out for VMs too
[10:26] <dmarkey> cool, sounds good
[10:28] <cjwatson> yay, I love moderately routinely having a two-step GPG verification path to upstream developers
[10:29] <dmarkey> so, if i were to suggest a PAE enabled netboot installer(possibly especially for Xen), where would i make such a suggestion
[10:31] <cjwatson> a bug on the debian-installer package in Ubuntu would be fine
[10:31] <dmarkey> so, in essence, you get to make the decision?
[10:31] <dmarkey> :)
[10:32] <cjwatson> he who does the work ...
[10:32] <cjwatson> I mean, you could bring it up at UDS if you want something more formal, but it doesn't seem big enough for that
[10:33] <cjwatson> and by big I mean complicated and requiring substantial design
[10:33] <dmarkey> i see
[10:34] <cjwatson> wishlist bugs are fine for small features
[10:35] <dmarkey> but it does mean for every release someone will have to roll this initrd/vmlinuz especially
[10:36] <cjwatson> no?
[10:37] <cjwatson> if we did it, we'd do so by changing the automatic build system
[10:38] <dmarkey> oh yea, silly me
[10:39] <dmarkey> where would i get my hands on a current lucid netboot image
[10:45] <cjwatson> http://cdimage.ubuntu.com/netboot/lucid/
[10:45] <dmarkey> great, im going to give it a whirl
[10:52] <dmarkey> https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/532547
[10:52] <ubottu> Launchpad bug 532547 in debian-installer "Create PAE enabled 32bit install vmlinuz/initrd" [Undecided,New]
[10:56] <michaelforrest1> ev: I guess I'll have a look at this map bug now - i.e. I'm going to walk over and bother Ken about it :)
[10:56] <ev> cool, thanks
[11:09] <ev> michaelforrest1: I tried to get suggestions from the platform team on a correct approach for dealing with it, but came up empty
[11:13] <michaelforrest1> ev: yeah that's been a very contentious area over the last couple of days
[11:24] <dmarkey> cjwatson: i'm having trouble getting the full set of module for that particular initrd
[11:41] <ev> cjwatson: is this reasonable? http://paste.ubuntu.com/388902/  I don't see how we can do any more, given the mentioned restraints that debconf puts on us.
[11:44] <cjwatson> ev: seems ok
[11:46] <ev> okay, cool
[11:50] <CIA-3> user-setup: evand * r216 ubuntu/ (3 files in 2 dirs): Provide a progress message for wiping swap space (LP: #432422).
[12:02] <ara> ev, in the ubiquity 2.1.31 changelog you mention fixing a bug where the first screen appeared before the translation was made
[12:03] <ara> ev, I still get this bug, though (in 2.1.31)
[12:03]  * ev pulls down a CD and checks
[12:05] <dmarkey> cjwatson: sorry for being a bother, where could i get the full module set for the kernel i that netboot initrd, I tried http://ge.archive.ubuntu.com/ubuntu/pool/main/l/linux/linux-image-2.6.31-14-generic_2.6.31-14.48_amd64.deb but the module didnt insert properly
[12:14] <cjwatson> dmarkey: well that should be the correct set
[12:14] <cjwatson> oh
[12:14] <cjwatson> 2.6.31 != 2.6.32
[12:14] <cjwatson> you might like to try 2.6.32-14 instead ;-)
[12:15] <cjwatson> (are you in Georgia?)
[12:15] <dmarkey> jesus, im stupid
[12:26] <ev> ara: can you elaborate on what you're seeing?  Do you briefly see text like ${MEDIUM} and ${RELEASE}?  Or is it just that you see English text when you're expecting something else?
[12:28] <dmarkey> cjwatson: hey, i see that with LVM, /boot is still sda5 :(
[12:29] <cjwatson> that bug's somewhere on my list
[12:31] <dmarkey> dont think i raised a bug for that particular issue
[12:33] <cjwatson> somebody else either did, or mailed me about it
[12:33] <ara> ev, the variables
[12:33] <ev> weird
[12:34] <ara> ev, do you see it in English? the last time I saw it, it was installing Xubuntu in SPanish
[12:34] <dmarkey> cjwatson: cool.
[12:35] <ev> ara: from the desktop, or booted straight into the installer?
[12:35] <ara> ev, straight into the installer
[12:36] <ara> ev, sorry, the last time it was Xubuntu, but in English, straight into the installer
[12:40] <ev> hrm, can't reproduce it with an Ubuntu CD.
[12:40] <ev> in English or Spanish
[12:48] <CIA-3> console-setup: cjwatson * r131 ubuntu/debian/ (5 files):
[12:48] <CIA-3> console-setup: Move to a single Upstart job (since ACMs and fonts are saved in /etc
[12:48] <CIA-3> console-setup: nowadays so there's no reason we can't set up both keymap and font
[12:48] <CIA-3> console-setup: early), set up to run just before Plymouth shows the splash screen
[12:48] <CIA-3> console-setup: (LP: #524439).
[13:22] <ev> ara: can you walk me through the exact set of steps you're taking to reproduce this (boot CD, select Spanish from the menu, move down to the Install Ubuntu option...)
[13:22] <ev> I was unable to reproduce it with the xubuntu CD either
[13:22] <ara> ev, I chose OEM (maybe unrelated, but I just remembered)
[13:23] <ara> ev, I will try to reproduce it again
[13:26] <ev> thanks
[13:26]  * ev lunch
[13:27] <ara> ev, I am still able to reproduce it. with the OEM option. I will try without
[13:31] <dmarkey> cjwatson: hey im going to raise a bug about hvc0 detection, that ok?
[13:33] <cjwatson> dmarkey: sure
[13:34] <ara> ev, mmm, I doubled checked the version of ubiquity and it says 2.1.30 :( sorry for confusion
[13:37] <ev> ara: no worries, I'm just happy that it's not a hard to resolve bug :)
[13:47] <dmarkey> cjwatson: https://bugs.launchpad.net/ubuntu/+bug/532651
[13:47] <ubottu> Launchpad bug 532651 in ubuntu "hvc0 is not detected/set up as a login tty(Xen Guest)" [Undecided,New]
[13:58] <dmarkey> cjwatson: i could try to investigate a patch if you wish
[14:19] <cjwatson> dmarkey: please attach /var/log/installer/syslog to the bug
[14:20] <cjwatson> dmarkey: also does /sys/bus/xen exist?
[14:20] <cjwatson> feel free to look into a patch if you like
[14:21] <cjwatson> ev: what's the status of the last two work items in foundations-lucid-oem-config?  are they going to make beta?
[14:24] <ev> I've yet to find any problems with non-ascii passwords, but I'll hunt down whomever came up with that item in the first place.  I'll endeavor to get the remove language pack removal done for beta.
[14:24] <dmarkey> cjwatson: the xenfs module isnt included in the initrd i created
[14:24] <dmarkey> should i include it?
[14:25] <ev> Though note that we have oem-config/install-language-support now, which I think partly solves the problem.
[14:28] <cjwatson> dmarkey: I don't know the details of what's needed for /sys/bus/xen to exist
[14:28] <cjwatson> ev: that's only half of it, isn't it?  I thought there was a different WI for installing vs. removing
[14:29] <ev> sure, but if you're an OEM you only need one or the other, no?
[14:30] <cjwatson> true
[14:32] <cjwatson> ev: non-ascii passwords: my understanding (perhaps faulty) was that it broke samba mounts?
[14:32] <cjwatson> ev: slangasek might be a good person to ask
[14:33] <cjwatson> also that merge review is unreadable; it should be re-prepared against lp:~ubuntu-core-dev/user-setup/ubuntu and then it might be possible to see the diff ...
[14:34]  * cjwatson comments on the bug to that effect
[14:34] <Omahn_> Is it possible to preseed the default OS that grub boots into following a preseed install?
=== Omahn_ is now known as Omahn
[14:35] <cjwatson> Omahn: only by a late_command that edits /target/etc/default/grub and runs in-target update-grub
[14:36] <Omahn> cjwatson: Thanks, I'll give that a look. Resizing ntfs during early_command with ntfsresize & parted --script commands works a treat.
[14:55] <CIA-3> kickseed: cjwatson * r275 ubuntu/ (debian/changelog kickseed.sh):
[14:55] <CIA-3> kickseed: Preseed base-installer/kernel/backports-modules to 'nouveau' if
[14:55] <CIA-3> kickseed: installing a desktop task, as a partial workaround for problems
[14:55] <CIA-3> kickseed: elsewhere (LP: #526422).
[15:01] <CIA-3> kickseed: cjwatson * r276 ubuntu/ (debian/changelog kickseed.sh):
[15:01] <CIA-3> kickseed: Be lenient if the Kickstart file author forgets the required space after
[15:01] <CIA-3> kickseed: '@' in a %packages section (LP: #419392).
[15:03] <CIA-3> kickseed: cjwatson * r277 ubuntu/kickseed.sh: move to right place
[15:04] <CIA-3> kickseed: cjwatson * r278 ubuntu/debian/changelog: releasing version 0.53ubuntu3
[15:07] <CIA-3> apt-setup: cjwatson * r183 ubuntu/ (debian/changelog generators/40cdrom):
[15:07] <CIA-3> apt-setup: Mount /proc, /sys, and /dev in the target system if necessary when
[15:07] <CIA-3> apt-setup: running apt-cdrom (LP: #527641).
[15:10] <CIA-3> apt-setup: cjwatson * r184 ubuntu/debian/changelog: releasing version 1:0.42ubuntu3
[16:03] <dmarkey> cjwatson: i included the xenfs module. is working now
[16:06] <dmarkey> so, im not sure if its a bug anymore
[16:18] <cjwatson> it's a legit bug if it doesn't work out of the box
[16:24] <dmarkey> maybe i'll just add xenfs to the list of modules to be included in my other 2 bugs
[16:26] <dmarkey> although i cant find any reference to xenfs in finish-install.d/*
[16:36] <CIA-3> ubiquity: cjwatson * r3876 ubiquity/ (debian/changelog scripts/install.py): merge lp:~didrocks/ubiquity/copy_wallpaper_cache
[16:55] <cjwatson> ev: do we need to add 'PROGRESS' to the list of questions monitored when calling user-setup-apply?
[16:56] <ev> it worked fine without it for me
[16:56] <cjwatson> ok
[16:56] <cjwatson> so is the ubiquity task on that bug invalid?
[16:57] <ev> I believe so, marking as such
[17:00] <cjwatson> thanks
[17:00] <ev> son of a...
[17:00] <ev> the fix I needed for the slideshow was sitting in its demo application all along
[17:26] <CIA-3> ubiquity: evand * r3877 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py):
[17:26] <CIA-3> ubiquity: WebKit puts file URLs in their own domain by default. This means that
[17:26] <CIA-3> ubiquity: anything which checks for the same origin, such as creating an
[17:26] <CIA-3> ubiquity: XMLHttpRequest, will fail unless this is disabled (LP: #532502).
[19:09]  * cjwatson commits the new gfxboot branding
[19:45] <CIA-3> usb-creator: superm1 * r270 usb-creator/ (debian/changelog usbcreator/frontends/gtk/frontend.py):
[19:45] <CIA-3> usb-creator: Move the populating of ISOs from ~/Downloads to before adding an ISO
[19:45] <CIA-3> usb-creator: from command line option --iso. This allows the command line fed
[19:45] <CIA-3> usb-creator: ISO to be selected automatically.
[19:48] <cody-somerville> Are there any plans to move to hybrid isos in lucid+1?
[19:48] <superm1> ev, do you know if r259.1.4 was just because of the hack behavior change in automatically, selecting or did poolie have some type of use case that needed it shown even with --iso?
[19:49] <superm1> i had intentionally added a commit before to hide it when --iso was added
[19:49] <cjwatson> cody-somerville: which of the several possible meanings of hybrid isos do you mean?
[19:49] <cody-somerville> cjwatson, http://syslinux.zytor.com/wiki/index.php/Doc/isolinux#HYBRID_CD-ROM.2FHARD_DISK_MODE
[19:51] <cjwatson> hmm, hadn't seen that.  it seems reasonably non-invasive
[19:51] <cjwatson> we'd have to finally get round to dealing with the syslinux/gfxboot merge though
[19:51] <cjwatson> please raise a bug on the ubuntu-cdimage project; it seems reasonable
[19:56] <cody-somerville> Filed LP #532880
[19:56] <ubottu> Launchpad bug 532880 in ubuntu-cdimage "Please add support for hyrbrid ISOs" [Undecided,New] https://launchpad.net/bugs/532880
[19:58] <mortal_> https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/528171 is this fixed?
[19:58] <ubottu> Launchpad bug 528171 in ubiquity "Lucid Alpha 3 installer crashes on manual partitioning" [Undecided,New]
[19:59] <shtylman_> mortal_: not to my knowledge
[19:59] <mortal_> I have a separate /home partition so I can not use the whole disk option
[19:59] <mortal_> and my karmic->lucid upgrade stalls with some waiting [SM] / message
[20:00] <shtylman_> mortal_: you can format the drive manually with fdisk or qparted
[20:00] <shtylman_> qparted might be on the live cd
[20:00] <shtylman_> but I don't remember
[20:00] <mortal_> what about cfdisk?
[20:00] <mortal_> I can mount the partitions manually
[20:00] <shtylman_> dunno what that is
[20:00] <mortal_> but where and how do I proceed
[20:00] <mortal_> the partitions are there
[20:00] <shtylman_> you don't need to mount them to install
[20:01] <mortal_> how do I do it then
[20:01] <shtylman_> ahh I see... well, you could try a newer cd
[20:01] <mortal_> downloading the latest now
[20:01] <shtylman_> like one of the daylies
[20:01] <mortal_> yup
[20:01] <mortal_> that one
[20:01] <shtylman_> also... the alternative cd might be an option
[20:03] <mortal_> downloading the alternative cd
[20:04] <mortal_> thanks
[20:06] <shtylman_> np
[20:08] <mortal_> I have had boot problems with all the upgrades
[20:10] <CIA-3> ubiquity: cjwatson * r3878 ubiquity/ (debian/changelog ubiquity/frontend/kde_components/PartMan.py):
[20:10] <CIA-3> ubiquity: * KDE frontend:
[20:10] <CIA-3> ubiquity:  - Fix references to PARTITION_* in PartMan module (LP: #528171).
[20:10] <cjwatson> mortal_: thanks for the note; your bug will be fixed in the next ubiquity upload
[20:11] <cjwatson> which should be in the next day or two
[20:34] <mortal_> the alternate installer is working ok at the moment but it is strobing weirdly
[20:35] <mortal_> on intel dg45fc board
[20:35] <mortal_> I could get some epileptic seizure :b
[20:37] <mortal_> ahh, my head
[20:37] <mortal_> *falls and has violent convulsions* nooo
[20:38] <mortal_> the graphics on my intel board are a bit weird
[20:44] <mortal_> the alternate install failed
[20:44] <mortal_> could not find some package, red screen
[20:44] <mortal_> so I will wait for the upload
[20:48] <mortal_> now downloading the alpha3 alternate
[20:48] <mortal_> let's see
[21:03] <cjwatson> dailies often fail with problems with individual packages
[21:03] <cjwatson> it's one of the reasons they're dailies not releases
[21:58] <superm1> cjwatson, how keen would you be on a hack like this http://pastebin.com/nVsQgCTs to work around that problem from the other day now that warnings are getting pushed up through to the frontend w/ recent partman-base?
[21:59] <cjwatson> superm1: I don't want to touch that until we actually understand the problem, and preferably have upgraded to parted 2.2
[21:59] <cjwatson> I absolutely don't want to just provide a switch to ignore errors
[21:59] <cjwatson> at least not until I know there's no alternative
[21:59] <superm1> cjwatson, OK.  well from what i can gather it started from ubiquity 2.1.28 being upgraded to 2.1.29, which most likely was from your partman-base fixes to let IS_CHANGED work
[22:00] <superm1> i can reproduce it on the 03/02 media by just upgrading ubiquity* and nothing else.
[22:00] <cjwatson> that would have been the upgrade to parted 2.1.
[22:00] <cjwatson> the old ubiquity linked against libparted1.8
[22:00] <cjwatson> the old parted_server, that is
[22:01] <cjwatson> I haven't been following.  what's the bug number?
[22:01] <superm1> i don't have a bug number filed for this, i really should
[22:01] <superm1> i'll get what i know thus far into one
[22:02] <cjwatson> from the circumstances, it's very likely either a parted bug or a failure of partman to handle something in parted 2.1 - either way I'd like to address that
[22:02] <superm1> well according to a diff between manifests, no changes to parted or libparted from 02->03
[22:03] <superm1> so parted 2.1 was already in the 02 builds
[22:07] <cjwatson> it doesn't matter whether parted 2.1 was there - ubiquity 2.1.28 was linked against libparted1.8.  https://edge.launchpad.net/ubuntu/lucid/i386/ubiquity/2.1.28
[22:07] <cjwatson> it wouldn't suddenly decide to use parted 2.1, because libparted-2.1 has a different soname
[22:08] <superm1> Oh i didn't realize that it was actually linking to it
[22:08] <cjwatson> ubiquity ships parted_server, which is a C binary
[22:08] <cjwatson> from partman-base
[22:16] <superm1> Ok, i've added relevant bits to bug 532961
[22:16] <ubottu> Launchpad bug 532961 in ubiquity "Unable to install from a recovery partition on ubiquity 2.1.29 and later" [Undecided,New] https://launchpad.net/bugs/532961
[22:19] <cjwatson> superm1: so there's an entry in parted 2.2's NEWS file describing what sounds like a similar problem, which is one reason I'm tempted to ascribe it to that
[22:20] <cjwatson> I've committed a merge of parted 2.2 to Debian's git repository, but I'm awaiting review by one of the other maintainers
[22:21] <cr3> I just nocied "verifying installation configuration" seems to show 800%, known problem or worth a bug?
[22:22] <cjwatson> cr3: worth a bug
[22:22] <cjwatson> superm1: parted 2.1 switched from using BLKPG to using BLKRRPART (actually sort of an older interface, but apparently more suitable for other reasons), which is exactly the sort of thing that could cause this
[22:23] <cjwatson>         linux-commit: do not unnecessarily open partition device nodes
[22:23] <cjwatson>         After patching parted with my do-not-use-BLKPG patch, I started
[22:23] <cjwatson>         to get EBUSY errors on commit_to_os. Note this is not caused
[22:23] <cjwatson>         by the do-not-use-BLKPG patch, this was already happening, but
[22:23] <cjwatson>         parted was silently ignoring the errors (and the kernel was
[22:23] <cjwatson>         not notified of the changes, which is bad).  The error now
[22:23] <cjwatson>         actually gets reported.
[22:24] <cjwatson> the commit message is a bit of an essay - read it for all you might ever want to know ;-)
[22:24] <superm1> Ah.  I'll grab parted 2.2 from debian git then and give it a shot
[22:24] <superm1> but that does sound quite much like it can be this situation
[22:26] <cjwatson> actually that may be the change that *introduced* the bug, but it's certainly related.  sadly the workaround in 2.2 is a crappy sleep
[22:26] <cjwatson> 2.2 is another soname change, so will be some work to introduce
[22:26] <superm1> probably need to rebuild ubiquity against it then because of parted_server?
[22:27] <cr3> cjwatson: done, reported bug #532970
[22:27] <ubottu> Launchpad bug 532970 in ubiquity "Verifying installation configuration seems to show up to 800% completion" [Undecided,New] https://launchpad.net/bugs/532970
[22:27] <superm1> or I can just temporarily use some symlink hacks just for a  tests sake
[22:27] <cjwatson> http://git.debian.org/?p=parted/parted.git;a=commit;h=ad25892bb995f61b0ddf801ed1f74e0b1e7390ce, http://git.debian.org/?p=parted/parted.git;a=commit;h=1d8f9bece138e4d8e58f7b059b4195aff6f39deb, http://git.debian.org/?p=parted/parted.git;a=commit;h=2a6936fab4d4499a4b812dd330d3db50549029e0, http://git.debian.org/?p=parted/parted.git;a=commit;h=0a21f0b7ed7ff0e536a5c30dfe1910c33d2ca243
[22:27] <cjwatson> I have no idea whether libparted will work if you lie to the linker
[22:28] <cr3> anyone happen to know where scripts/init-top in the initrd happens to land in a live environment?
[22:28] <cjwatson> well it's the other way round, scripts don't get copied from the initrd into the live environment, the initrd gets built from the filesystem that becomes the live environment
[22:28] <superm1> wow that last commit really is ugly
[22:28] <cjwatson> but /usr/share/initramfs-tools/scripts/init-top/ anyway
[22:29] <cjwatson> yeah, almost certainly needs more thought than that!
[22:30] <cjwatson> what's weird is that I already have an Ubuntu patch in there that calls udevadm settle
[22:32] <cjwatson> superm1: I don't suppose you could arrange to have 'udevadm monitor' running across the failure?
[22:33] <superm1> cjwatson, shouldn't be a problem
[22:33] <cjwatson> that might offer a hint (or might perturb away the problem - it could be racy)
[22:36] <cr3> cjwatson: by the way, I'd like your advice on getting bootchart from the installer environment. I noticed that keybuk does some munging of the initrd.lz file prior to the netinstall, would it be preferable to contain bootchart stuff (scripts and configuration) in a udeb instead?
[22:37] <cjwatson> the live environment doesn't use udebs
[22:37] <cr3> cjwatson: heh, that solves it then :)
[22:38] <cjwatson> just install bootchart in a live environment (or a copy of it with suitable virtual filesystems mounted) and regenerate the initramfs
[22:39] <superm1> http://launchpadlibrarian.net/40283275/udevadm_monitor.txt
[22:39] <cjwatson> you'll get an initrd.gz out instead of an initrd.lz, but that doesn't matter; you can recompress it if you like but the reason we use .lz is just to save some CD space.  if you use initrd.gz, be careful to adjust isolinux.cfg to match
[22:39] <cjwatson> superm1: could you attach matching partman and syslog for that run so that I can correlate timestamps?
[22:40] <cjwatson> (albeit probably not tonight)
[22:40] <cjwatson> hm, not a whole lot happening there, just change events
[22:40] <cjwatson> any non-default udev rules in this environment, just so I know?
[22:42] <superm1> there is a rule to hide recovery partitions from devkit/udisks
[22:43] <superm1> http://launchpadlibrarian.net/40283482/syslog and http://launchpadlibrarian.net/40283493/partman are from that same failed run with udevadm monitor running
[22:44] <superm1> there are some prep steps that happen before ubiquity gets started for setting the recovery partition active and loading up grub onto it just in case there is a failure during the install (you'll see them in syslog too)
[22:44] <CIA-3> ubiquity: evand * r3879 ubiquity/ (bin/ubiquity-dm debian/changelog):
[22:44] <CIA-3> ubiquity: Restore Colin's set_locale function. However, this is only half the
[22:44] <CIA-3> ubiquity: puzzle. We need to finish running the language component when the
[22:44] <CIA-3> ubiquity: exit button is pressed when in greeter mode.
[23:19] <CIA-3> ubiquity: cjwatson * r3880 ubiquity/ (bin/ubiquity-dm debian/changelog): ubiquity-dm: Close stdout from dbus-launch.
[23:21] <CIA-3> ubiquity: cjwatson * r3881 ubiquity/ (7 files in 4 dirs): autogen
[23:25] <CIA-3> ubiquity: cjwatson * r3882 ubiquity/ (24 files in 9 dirs):
[23:25] <CIA-3> ubiquity: Drop real UID/GID when initialising KApplication. The real and
[23:25] <CIA-3> ubiquity: effective IDs need to match, but raising them means we can't talk to
[23:25] <CIA-3> ubiquity: D-Bus. Instead, use saved IDs to store our root privileges so that we
[23:25] <CIA-3> ubiquity: can get them back (LP: #526486).
[23:37] <CIA-3> ubiquity: cjwatson * r3883 ubiquity/debian/control: whoops, let's update d-i so that I get a current libparted build-dep
[23:39] <CIA-3> ubiquity: cjwatson * r3884 ubiquity/ (d-i/manifest debian/changelog):
[23:39] <CIA-3> ubiquity: Automatic update of included source packages: apt-setup 1:0.42ubuntu3,
[23:39] <CIA-3> ubiquity: user-setup 1.28ubuntu4.
=== robbiew is now known as robbiew_