How do you make sure that your passwords are safe? You can make them longer, complicate them by adding odd characters, making sure to use different passwords for each user account that you have. Or, you can simply skip them all together.
The secure shell, ssh, is a key tool in any Linux user's toolbox. As soon as you have more than one machine to interact with, ssh is the obvious choice.
When logging into a remote machine through ssh, you are usually prompted with the remote user's password. An alternative to this is to use an asymmetric key pair.
An asymmetric key pair consists of a private and public key. These are generated from an algorithm - either RSA or DSA. RSA has been around for a long time and is widely supported, even by old legacy implementations. DSA is safer, but requires v2 of the ssh protocol. This is not much of an issue in an open source world - keeping the ssh daemon implementation up to date is not a problem, but rather a requirement. Thus, DSA is the recommended choice, unless you have any specific reason to pick RSA.
The generated keys are larger than a common user password. RSA keys are at least 768 bits, default 2048 bits. DSA keys are 1024, as the standard specifies this.
To generate a DSA key, use the following command:
$ ssh-keygen -t dsa
This generates the files ~/.ssh/id_dsa and ~/.ssh/id_dsa.pub. You can specify a passphrase in the key generation process. This means that they key only can be used in combination with a passphrase, adding to the security.
The generated file ending with pub is the public half of the pair. This can be shared with remote hosts that you want to ssh to. The other file, id_dsa, is the private half of the pair. This must be protected just as you password. I.e. do not mail this, do not store it on untrusted machines, etc.
Having a 1024 bits long key can be thought of as having a 128 characters long password. This means that the key pair method is safer than most passwords that you can remember. Keys are also completely random, so they cannot be cracked using dictionary attacks. This means that you can increase the safety of your remote host by disabling logins using passwords, thus forcing all users to use key pairs.
Having generated your key pair, all that is left is copying the public half of the key to the remote machine. You do that using the ssh-copy-id command.
$ ssh-copy-id -i ~/.ssh/id_dsa.pub user@remote
This adds your key to the remote machine's list of authorized keys. Just to be on the safe side, it is also good to ensure that the ~/.ssh and ~/.ssh/authorized_keys aren't writable by any other user than you. You might have to fix this using chmod.
Having added the key to the remote machine, you should now be able to ssh to it without using a password.
$ ssh user@remote
This applies to all sshd-based mechanisms. So you can scp freely, as well as mount parts of the remote file system using sshfs.
One potential catch twenty two issue here is if the remote machine does not allow password-based logins. Then the ssh-copy-id command will not work. Instead you will have to take the contents of the public key half and manually add it as a new line to the ~/.ssh/authorized_keys file on the remote machine. This is what the ssd-copy-id command does for you.
This also tells you what to do if a key is compromised, or simply falls into disuse. Simply remove the corresponding line from the remote's list of authorized keys. You can usually recognize the key in question from the end of the line where it reads username@hostname.
So, until next time, no more passwords!