Re: best practices: public key authentication

Chuck wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Nico Kadel-Garcia wrote:
| Chuck wrote:
|> -----BEGIN PGP SIGNED MESSAGE-----
|> Hash: SHA1
|>
|> I'm curious to find out what others think about pubkey authentication
|> best practices. Assuming your private key is protected with a strong
|> passphrase, is there any value in occasionally regenerating your keypair
|> and replacing your public key on servers that you use pubkey
|> authentication with?
|
| Yes. It helps prevent fascinating man-in-the-middle attacks if you use a
| public key for multiple remote targets, and it reminds you and others to
| discard access you don't need any longer.

What exactly is a "fascinating" man in the middle attack and how does
changing my keypair prevent it? I thought MITM attacks would be detected
by the server's key changing. OpenSSH (and presumably others) warn you
if the key of the server does not match what you've previously known it
to be.

Many SSH servers have poor local security: sharing home direcotories via NFS, for example. Many SSH servers also have poor control over their hostbased private keys. With these, a man-in-the-middle attacker can theoretically pretend to be your SSH target, allow you to log in without password and especially with key-forwarding enabled, and allow you to connect to their man-in-the-middle, securely. They then provide an unsecured monitor in the middle, and pass along *ANOTHER* secured transaction to your destination host.

Most of us would never notice such an attack. A few such leveraged attacks on a source code repository could create all sorts of problems for a secure source code repository.
.

Relevant Pages