Hi,

Can someone step me through the login process when using Kerberos authentication?

eg.
1) user enters username/paassword
2) server looks for local account (or kerberos searches AD for username)
3) etc

I am particularly interested in the root login, if it differs from a normal user - as I had an issue not being able to login as root after a power outage (with AD temporarily unavailable).

Thanks.

Well, actually kerberos works differently from what you have described above. Kerberos is a secret key authentication mechanism, whereby the client and the server hold a secret symmetric key (ticket) for some previously negotiated encryption algorithm. This link explains it in more detail: http://web.duke.edu/~rob/kerberos/kerbdetails.html. The exact spec of how AD deals with kerberos I am not sure of, but i do know you can setup AD clients to 'cache' recently authenticated sessions for exactly ther reason you talk about:

I guess my question was more specifically about the PAM login process when using Kerberos authentication to AD.

My understanding is that the first line in this PAM file initiates the Kerberos process:

account sufficient /lib/security/$ISA/pam_krb5.so
account required /lib/security/$ISA/pam_unix.so

Which appears to show that for all local accounts, including root, an initial attempt will be made to authenticate with AD. Only then will the Kerberos process itself kick in (if authentication is successful).

I might try caching login details, but would prefer root to bypass the process above.

OK, after some testing it appears that a root lockout problem I was having may not have been directly related to Kerberos authentication.

I think I now understand how the PAM login process works.

No, the way this is laid out: what happens is PAM checks if a kerberos ticket exists or can be found for root. If not, it checks local unix/linux accounts. So if you can't log in as root, where root is a local account, it's nothing to do with AD, so far as I unserstand it. YMMV.