Should I change the letters sdd to sdb, sdc, sda accordingly to the replaced HDD or is it always the same procedure (except fdisk)?
I'm not a tera guy, since you pm'd me I'll take a shot. But take what I say with a grain of salt.
But looking at the relavent pieces of that DOC md1, md0 and md2 are all separate arrays/devices.
MD0 is /boot where the kernel uImage is probably kept
MD1 is / where the rootfs (linux OS) is kept
MD2 is /mnt/array1 where the data array is kept
The document you reference uses the "natural" ability of the system to rebuild the raid by swapping out and in disks one by one. This way you would not have to create the raid(s) and place the kernel and rootfs in the right places yourself. They would be recreated/repopulated automatically.
There is probably a mdadm config file somewhere that is probably kept at /etc/mdadm.conf
that is keeping track of how the arrays are built. However you would need to keep track of the device you are inserting to the box. And change the letter accordingly as you said when readding it to the array.
Take a look here as well... There is some good info on this wiki I think.http://linux-raid.osdl.org/index.php/Growing