Comment 18 for bug 759545

Regarding comment #13 -- The error messages are cosmetic. It is caused by the "virtual" partitions in EC2, where a raw disk is presented as a /dev/{s,xv}da1 when /etc/kernel/*/zz-update-grub is executed. The kernel update script that handles the kernel updates on EC2 is /etc/kernel/*/x-grub-legacy-ec2. On both instance-store and EBS instances, there is no need for to run zz-update-grub, since the paravirtual grub (pvgrub) is being run outside the instance. However for HVM instances, we are using Grub2 to boot the images.

Essentially, the scripts should be mutually exclusive. In order to fix this problem, there needs to be a mechanism for deciding which one should run -- for example if it is running on bare metal or HVM, then it would run the zz-update-grub, but if it was running on EC2-like hardware, it would run x-grub-legacy-ec2.