MDADM RAID Failed After Upgrade from Ubuntu 8.10 to 9.04

With no surprise the update from Interpid Ibex to Jaunty Jackalope was not as easy as I hoped. After booting the fresh 2.6.28 kernel I was welcomed with “Press enter for maintenance (or type Control-D to continue)” and errors when trying to mount my RAID0. After some investigations found that the devices are now named md_d* and there are some devices in /dev/md/.  Because of this change I had to do the following things to get my raid up and running:

  1. Reassemble the raid (check if it is inactive with less /proc/mdstat) – in my case the command looked like this: mdadm --assemble /dev/md_d1 /dev/sdb2 /dev/sda5
  2. mdadm --detail --scan >> /etc/mdadm/mdadm.conf and comment your previous ARRAY definition in the config file
  3. Modify /etc/fstab if needed

Also if DKMS fails to build the new NVIDIA driver be sure to check that you have installed the headers files (the upgrade process did not install them for me).

Again : I’m NOT responsible for any damage or data loss you might experience. Backup often, backup soon!

How-to Install Nvidia Driver on 2.6.25-2 Debian Kernel (with Xen)

Another big surprise rolled out from the Debian Unstable repository when today I updated my kernel from 2.6.25-1 to 2.6.25-2 – the maintainers have enabled Xen in the non-Xen kernel (linux-image-2.6.25-2-686). The result from this is that Nvidia drivers don’t compile throwing error about Xen enabled kernel.

The solution (if you don’t want to recompile your own kernel) :

As usual backup before executing any of these :)

  1. cd /lib/modules/2.6.25-2-686 (or wherever are your modules)
  2. grep CONFIG_XEN . -Rl | xargs -i sed -i -e ‘s/CONFIG_XEN/CONFIG_LUPO/g’ {}
  3. export IGNORE_XEN_PRESENCE=1
  4. sh NVIDIA-Linux-x86-173.08-pkg1.run (or you can use patched 169.12 – http://www.nvnews.net/vbulletin/showthread.php?t=110088 )

Source: linux-image-2.6.25-2-686: nVidia driver fails to compile due to active Xen support