Repairing a Debian system that no longer boots after upgrades
Summary
Manuel Schmitt documents a rescue procedure for a Debian system whose upgrade left the kernel package, GRUB configuration and mdadm configuration inconsistent. The affected system had been upgraded from Debian Lenny to Squeeze. /boot contained a bzImage, but not the vmlinuz that GRUB detection expected.
Ideas
- An empty grub.cfg can result when no suitable linux-image is installed.
- A rescue system with bind mounts allows package and bootloader repairs in a chroot.
- mdadm --examine --scan can reconstruct a lost RAID configuration from existing metadata.
Insights
- Regular package updates do not guarantee a complete boot path across several distribution upgrades.
- Bootability has to be tested explicitly after kernel and RAID changes.
Facts
- The repair reinstalled a linux-image package in the chroot.
Recommendations
- Before a distribution upgrade, check the kernel package, /boot usage and RAID state.
- Keep a tested rescue system and a documented chroot procedure ready.
References
Read the original article on Hostblogger
Links to the original source and the Web Archive open in a new tab.