Information Security Stack Exchange is a question and answer site for information security professionals. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

It seems that they are mutually exclusive, as disabling one gives me the other, and vice versa. Two-factor auth for my ssh servers sounds really nice, so is there any way to accomplish this?

share|improve this question
Are you are not wanting to count passphrased ssh keys? – Phillip Nordwall Jul 31 '12 at 17:57
Oh, right. I should have specified that. No, that doesn't count. I'd like the server to have to be authenticated against twice, not the client :-) – Chris Blake Jul 31 '12 at 18:07
@ChrisBlake - why? What problem are you trying to solve? Can you be more specific? What's your threat model? What risk are you trying to defend against? – D.W. Aug 1 '12 at 2:13
@D.W. Threat model for requiring this: working with people you don't trust to take security as seriously as you do. You want to make it impossible for someone to compromise your server if their laptop, with a carelessly unencrypted ssh key, is stolen. – user18203 Dec 30 '12 at 4:19
@JaneDoe, if that's the problem, then this might be better solved through policy rather than a technical mechanism. Requiring a password on every login has major disadvantages, and it sounds counterproductive to me. I think it's better to just set an organizational policy requiring all your sysadmins to encrypt their private key with a passphrase. (If you don't trust your sysadmins to follow this policy, why are you letting them administer your systems?) – D.W. Dec 31 '12 at 22:33
up vote 23 down vote accepted

With recent Fedora and RHEL 6 releases, you can use RequiredAuthentications2 pubkey,password to require both pubkey and password authentication. Usually this is done to require pubkey and 2-factor authentication token, not the user's password.

Update: Now on RHEL / CentOS 7, and any system with a recent version of OpenSSH, you can use:

AuthenticationMethods "publickey,password" "publickey,keyboard-interactive"

It's also possible to use the Match directive to exclude IPs or Users.

share|improve this answer
With these new releases, this is the new (and easiest) correct answer. – Chris Blake May 18 '14 at 4:13
See @Jakuje answer here for more recent Linux distro key+password SSH setup. The name of config parameter has changed. – Mister_Tom Feb 18 at 22:56

You can have both public-key and password authentication on the same server. If public-key authentication fails, it will go to password authentication.

As to requiring both, that's seems silly and counterproductive, and checking man sshd_config there isn't an option to do this.

Your ssh private key should have a secure passphrase. So if an attacker obtains your private key, they still can't do anything without first obtaining your passphrase. If they've compromised that passphrase (most likely with a keylogger; or from brute forcing an extremely weak passphrase) they can trivially also grab/brute force any memorized password.

If you really want, you could possibly setup something with say ForceCommand (e.g., only allow public-key authentication and then direct user to a shell that prompts for a password). I don't recommend this.

A better alternative if you want to limit exposure, is to have a firewall setup to limit IPs that can reach the ssh port; possibly with an additional VPN running on a server somewhere if you may need to tunnel from another computer at some point. You could also use something like knockd to open a hole in a firewall after a particular port-knocking pattern, though recognize that anyone eavesdropping on traffic could replay the knocking pattern to open up a port.

share|improve this answer
Right. I should have taken the hint from @Phillip's comment, since the private key already needs a password, it's already a form of two-factor auth. And yeah, I'm also going to set up a port knocking daemon, but I couldn't choose from the given implementations on the website. Thanks for the referral and answer. – Chris Blake Jul 31 '12 at 18:39
Consider the case when the attacker has taken control of a machine in which the user has used an authentication agent (e.g., ssh-agent) to cache the private key. The attacker doesn't have the passphrase, but can still use the private key until the agent flushes its cache. It's in this scenario that requiring the key and the login password can help. – Psychonaut Jun 22 at 6:33

(cross posting SO answer with updated solution to these days)

If you read through the manual page for sshd_config(5), there is option AuthenticationMethods, which takes the list of methods you need to pass before you are granted access. Your required setup is:

AuthenticationMethods publickey,password

This method should work all the current Linux systems with recent openssh (openssh-6, openssh-7).

Older systems

The only exception I know about is RHEL 6 (openssh-5.3), which requires setting different option with same values (as described in the other answer):

RequiredAuthentications2 publickey,password
share|improve this answer
in "AuthenticationMethods", Instead of "pubkey" should be "publickey". Check: man sshd_config – lepe Jul 2 at 3:38

I looked into this a little more and came up with the following.

You could use PAM for two factor authenticaion, but in doing so you won't be using SSH keys you will be using a different two factors.

For example, You could use google with their two factor authentication and use pam to authenticate, as described at

share|improve this answer
can we use SSHkeys AND 2FA ? – confiq Sep 9 '15 at 13:01
Using a passphrased key gives you two factors as it is something you have and something you know. – Phillip Nordwall Sep 13 '15 at 16:22

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.