Upon encountering a checksum mismatch we could add bitflip detection to give us signal around faulty hardware. In theory this could be computationally very expensive so some basic performance testing will be required to determine if this is something we could run in production.

The algorithm to detect the bitflip would be as follows:

• For every bit in the disk image of the block
• Flip the bit
• Recompute the checksum
• Check if it matches the expected checksum