initramfs-tools failed to install/upgrade, due to not being able to write initrd or vmlinuz.

The important bit here is apparently:

  Setting up initramfs-tools (0.92bubuntu71) ...
  update-initramfs: deferring update (trigger activated)
  lzma: Encoder error: -2147467259

which simply means that there was insufficient space on the device to write the new kernel module.

However, I tried this again using a much larger USB device and generated an entirely different error - the update-initramfs trigger attempts to stat /vmlinuz and this a a link to a file that no longer exists, so again the wireless installation fails.

I created an Ubuntu 10.04 Beta 2 USB Startup Disk with 4GB persistent space and ran a full update after booting.

The link /vmlinuz points to a missing file boot/vmlinuz-2.6.32-19-generic - this file does not exist on the USB Startup Disk, it is not installed when the disk is created. The absence of this file prevents some updates and package installations from completing.

As a workaround, sudo cp /boot/vmlinuz-2.6.32-20-generic /boot/vmlinuz-2.6.32-19-generic, will allow installation to complete. There is an error on the first reboot, after which Ubuntu 10.04 appears to boot correctly with wireless enabled.

thanks for your report.

Maybe related to bug 557023

I see the same error when installing 10.04 on USB stick. It is curious that doing the same while using LiveCD session proceeds without any errors. The same media, same hw.

In my case ( bug #645700 ) there were over 700MB free, so presumably lzma did not run out of space.

In my case, /vmlinuz does point to a valid file, but I have no idea if this was true before the update:

~$ ls -al /vmlinuz
lrwxrwxrwx 1 root root 30 2010-09-17 00:28 /vmlinuz -> boot/vmlinuz-2.6.32-24-generic
~$ sudo ls -al /boot/vmlinuz-2.6.32-24-generic
-rw-r--r-- 1 root root 4034976 2010-09-16 18:14 /boot/vmlinuz-2.6.32-24-generic

This bug appears because on live systems, new initrds are written to /cdrom/casper/initrd.lz.new (see special treatment of live systems in /usr/sbin/update-initramfs starting from line 9).

If you create a persistent live system via Startup Disk Creator and pull the "How much" slider to the right as far as possible (what seems sensible for most cases and users), all space not used for the system is allocated for that documents-and-settings-ext2-file and there is no space left on /cdrom. Therefore update-initramfs fails with that cryptic lzma error code.

Perhaps Startup Disk Creator should leave some spare space on the live medium even if the user chooses "use all remaining space available for my live medium".

I think there are several things that need to be done to address this class of bugs:

1. Jockey needs to be more attentive to the lack of space on the root device. This will be the focus for bug #685017. It should detect the situation and give the user an error message dialog, without trying to do the installation.

2. usb-creator needs to not let the user reserve 100% of the space to documents, or should display an appropriate warning or UI to indicate that if they do not preserve some space for system files, the usb key will become non-functional. I recommend something like the following:

  * For 1G USB keys, do not offer persistent storage at all
  * For 2G to 4G USB keys, make sure there is 800M or so remaining after creating casper-rw - enough space to update to a new kernel, write initrd, and other system type things needed.
  * For >4G USB keys, put the casper-rw on its own partition separate from the system files. Ensure there's at least 1G of space for system files, to install kernels and so on.

Focusing this bug report on item #2 in my previous comment.

I think partly there may be some confusion in the UI. The slider doesn't really make it clear that in involves a tradeoff of space allocation between the system boot files, and the installation. So maybe it could use something that turns red or shows a warning if you drag above some threshold (90% or 800M or whatever).

With a 1G usb stick, it might make sense to advise the user that they will not be able to install kernel drivers due to space limitations.

It may be that the UI is a bit confusing, in that it's a bit non-intuitive that moving the slider all the way to the right could make the stick not work properly. The wording "persistence" could be interpreted to mean that it provides persistent space for system updates including kernel installs and so on, and people who would want to do that would logically choose to maximize the space for that, and be surprised that it didn't work this way.

An idea to make this clearer could be to present the slider as a choice between "document storage" and "system space", or some such. Or, it could show something red or pop a warning if the user selects an amount of space which could make it difficult to update kernel drivers. (Or maybe both.)

Actually, with a bit more testing, the issue is not that there isn't space but that the boot partition isn't writable, or at least writes don't persist from one boot to the next.

It is starting to sound like this is just a fundamental issue with the livecd environment, and that installation of -fglrx and -nvidia in livecds is just not a supportable option at this time. So I guess Jockey should be configured to not permit the user to do this (bug #685017).

I'm happy to add a reserved amount of space for the kernel and initramfs as a stop-gap measure until we have GRUB2 on the CDs. After which, I'd like us to write the kernel and initramfs in casper-rw, and have grub first attempt to read it from there.

This bug was fixed in the package usb-creator - 0.2.27

---------------
usb-creator (0.2.27) natty; urgency=low

  [ Mario Limonciello ]
  * If the EFI bootloader isn't present in the proper location but efi.img is
    available in boot/grub, extract the EFI bootloader and place it in the proper
    location. (LP: #677260)

  [ Evan Dandrea ]
  * Hide partition table block devices by default. To revert to the
    previous behavior, pass --show-all to usb-creator-gtk.
  * Increase the minimum persistent storage size to 1G.
  * Add a 30MB padding for the kernel and initramfs (LP: #562312).
 -- Evan Dandrea <email address hidden> Fri, 28 Jan 2011 15:42:25 +0000

I've seen quite a few apport-package bug reports coming in from people using USB persistance with Natty but having created them with Maverick which still has this bug. It seems like it'd be a good idea to fix this in Maverick and prevent reporting of these bugs.

Thank you for taking the time to report this bug and helping to make Ubuntu better. Unfortunately, the package installation error message that you have received is actually a failure with lzma and a generic one at that. The following steps may help resolve the issue. Please execute the following command, as it will clear your package cache, in a terminal:

sudo apt-get clean

Then try performing the update again. This may resolve your issue. We plan to make lzma errors more informative and that work is being tracked in bug 691985 which I will mark this a duplicate of.

This version has expired