I've found myself owning a few servers with large redundant disks andIf it is the same size as your current root partition, I'd mirror them
have started thinking about setting up a bootable 2G partition on the
secondary array as an alternate root partition for emergency
use/disaster recovery type applications. I have a basic idea of what I
(and the /boot partition too) using md raid1, and set grub to have the
option to boot either. That way if one disk goes down, you have a full
root system to come up to.
i have combination of both on our servers
all vital partitions are in raid1, i have two completely separate installations (well, they do share swap partitions)
raid1 won't protect you against file system corruption or operator's mistake/stupidity
both boot partitions have to be on the first and the second drive with raid-extra-boot=mbr-only so that both of them are bootable (and the same drives because map file has to be on the same drive as /boot partition is)
in addition, every change to fstab/lilo goes to both normal and emergency installations, so i can always reconstruct lilo from whichever installation i want
as for the tools -> just the basic stuff like gcc, lilo, mdadm, raiserfs tools, etc
whatever i need, i can always emerge
our servers are in the hosting centre located in another country, therefore it is important that networking is set up properly in both installations, so that i only need to ask hosting centre operators to reboot them into another one and i can immediately access them (if lilo -R doesn't work anymore)
just my 2/100 in the local currency
regards, martin
