Next: Garbage Collection Policy
Up: paranoid WHAT ON/OFF
Previous: paranoid retrieval ON/OFF
  Contents
When a file is to be added to the database, and its hash already exists,
retrieve the file in the database and do a byte-for-byte comparison with
the incoming file.
It's not clear what should happen if a hash collision does occur.
Here are some probably causes and remedies:
- The most probable cause of hash collision is some kind of data corruption,
usually due to hardware or software problems.
- Hash collision can also occur if hashes are truncated to too
short a length, or if a poor hash algorithm is used (MD5, SHA1, or a
similarly strong hash function capable of being used for cryptographic
authentication are recommended). In this case the hash length should be
increased, or the failing Hash function replaced. The entire database
will have to be rebuilt as a side-effect--see section 2.2.3.
- A hash collision between two full-length supposedly
`cryptographically strong' hashes is almost certainly a fatal failure
of the hashing algorithm, on which the entire fakefs database is
based (not to mention a number of cryptosystems, assuming well-known
cryptographic hash algorithms are used).
Next: Garbage Collection Policy
Up: paranoid WHAT ON/OFF
Previous: paranoid retrieval ON/OFF
  Contents
Zygo Blaxell
2003-03-04