In message <3C118DE4.1060900@mac.com>, you write: 

-> A little confusing, and definitely a problem. Is there a way to map the 
-> drives so that they're not going to move. If I disable the raid 
-> controller, my disk becomes wd0 again, which will fail during boot-up. I 
-> can see a situation making a reboot difficult with a bad disk on the 
-> raid controller.

What you can do is build a custom configuration where you `wire down' each
device.  What that means is that rather than having the following:

pciide* at pci* dev ? function ?
wd* at pciide? channel ? drive ?

You do:

pciiide0 at pci dev N function M
wd0 at pciide0 channel 0 drive 0
wd1 at pciide0 channel 0 drive 1
...

pciide1 at pci dev X function Y
wd2 at pciide1 channel 0 drive 0
wd3 at pciide1 channel 0 drive 1

Where N/M and X/Y are the device/function pairs for each of the IDE 
controllers in your system (which you should be able to find by booting
you current kernel and reading the `dmesg' output)... This should be 
resilient as long as the BIOS doesn't renumber devices when you turn
off the on-board IDE, for example (which I don't think any PCI-clued-
in BIOSes should do).

The syntax here may be slightly off as I wrote this from looking at my
dmesg output rather than my config file (the machine with my sources is
off right now), but the basic idea stands.

Obviously, this requres you to configure and build a custom kernel for
your machine as the GENERIC/INSTALL type kernels wildcard as much of
the configuration as possible to work on the broadest possible range
of system configurations.

--rafal

----
Rafal Boni                                                  rafal@mediaone.net