Rebuild raid concept

rebuildraid.py

#!/usr/bin/python -ttu
#
# For a set of disks that were previously in a RAID set, build an image of the data such as it was,
# ignoring parity.

class RAIDDEV:
  def __init__(self, DEVICE, OFFSET=0):
    self.DATA=file(DEVICE, r)
    self.DATA.seek(OFFSET)

# Set this to the byte offset of the start of data on *each* disk
OFFSET=0
# Make a list of each source device, in RAID order
DEVICES=(RAIDDEV('/dev/sda', OFFSET), RAIDDEV('/dev/sdb', OFFSET), RAIDDEV('/dev/sdc', OFFSET))
# This is ignored and assumed
RAIDLEVEL=5 
# Set the stripe size used in creating the original raid set
STRIPESIZE=(128 * 1024)
# Configure the output file
OUTFILE=file("/mnt/disk.img", "w")
# Assumes the first parity data is stored on the last disk on the first loop
XORPOS=len(DEVICES)-1

while True:
  for DEVICE in DEVICES:
    if DEVICES.index(DEVICE) == XORPOS:
      # Skip over the stripe containing redundancy
      next
    OUTFILE.write(DEVICE.DATA.read(STRIPESIZE))
    XORPOS+=1
    if (XORPOS > len(DEVICES):
        XORPOS=0
        
# Note: this program will exit on EOF and should instead exit on all disks being completed instead.
# As a result, N-1 stripes of data could be lost at the end of the disk.