Re: [bitfolk] Example of using the rescue environment to diagnose and fix a booting issue

On Sun, Feb 17, 2019 at 05:55:34PM +0000, Andy Smith wrote: Right, so the situation here was: - Customer had just done a do-release-upgrade from ubuntu 14.04 to 16.04 and then rebooted. - VPS did not boot and upon looking at the console it was at a GRuB screen saying: *GRUB-legacy from /boot/grub/menu.lst and then: Failed to boot both default and fallback entries. Press any key to continue... Based on that information I know: - /boot/grub/grub.cfg (modern grub config file) doesn't exist - /boot/grub/menu.lst (legacy grub config file) does exist but doesn't work These days we generally recommend installing grub-pc package to get the GRuB bootloader and letting it manage /boot/grub/grub.cfg (paths differ on CentOS). So anyway: - Connect to Xen Shell - Boot rescue environment - chroot into actual VPS - Discover /boot/grub/menu.lst that contains no kernels - See there actually are kernels in /boot - Try to update-grub but it isn't installed, implying grub isn't installed at all - apt install grub-pc - set some more sensible options in /etc/default/grub - update-grub - exit out of everything and boot the VPS - Success Hypothesis: grub-legacy was uninstalled during do-release-upgrade and never replaced by grub-pc leading to no working bootloader config. Cheers, Andy -- https://bitfolk.com/ -- No-nonsense VPS hosting