This problem occurred in both CentOS 5.2 and the new (awesome!) 5.3:
I'm using a SuperMicro motherboard with a 6 port NVidia SATA controller on the motherboard and an 8 port SuperMicro Marvel controller in a slot.
In my kickstart file, I have
part raid.01 . --ondisk=sda .
part raid.02 . --ondisk=sdb .
part raid.03 . --ondisk=sdc .
part raid.04 . --ondisk=sdd .
.
raid / --level=RAID1 --fstype=ext3 --device=md0 raid.01 raid.02 raid.03 raid.04
When I only have 4 drives installed in the system, this works fine.
However, whenever I have more than 4 drives, it seems that anaconda ignores my --ondisk, and lays out /dev/md0 somewhat randomly across the drives.
In particular, with 8 drives (sda..sdh), with the exact same kickstart file, it lays out /dev/md0 on drives sda, sdb, sdg and sdh.
with 10 drives (sda..sdj), with the exact same kickstart file, it lays out /dev/md0 on drives sda, sdh, sdi, sdj.
Does this make sense to anybody? Any help would be appreciated greatly.
Thanks!
Mark