Christoph Kaegi writes:
> Hi
> 
> What is the proper procedure to replace a defective SCSI hotswap
> disk with RAIDFrame?

While there are probably a few different ways, the one I've used most 
often is:

 1) Make sure all RAID components which live on the disk are marked 
as failed. (e.g. if there is a read error on just one part of a disk, 
and you have components on that disk for different RAID sets, you 
probably want to make sure they get marked as failed before removing 
the disk).  This step isn't strictly required, but it means that 
RAIDframe doesn't have to figure out that the component is gone.

 2) Pull out the defective component and put in the new one.

 3) If necessary, use scsictl to scan the bus.  (Some controllers 
need this, some don't.)

 4) Put a disklabel on the new disk.

 5) Use "rebuild-in-place" ('raidctl -R') for each RAID set to get
the new components re-synched with the rest of their respective 
RAID sets.

 6) Remember to fix up any boot blocks or other "magic" on the new 
disk.

The above has worked quite nicely on IBM x330 boxes w/ hot-swap U160 
SCSI...

Later...

Greg Oster