home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.kerberos
- Path: sparky!uunet!stanford.edu!OCFMAIL.OCF.LLNL.GOV!nessett
- From: nessett@OCFMAIL.OCF.LLNL.GOV (Danny Nessett)
- Subject: how does kerberos handle su utility
- Message-ID: <9209041608.AA16275@ocfmail.ocf.llnl.gov>
- Sender: news@shelby.stanford.edu (USENET News System)
- Organization: Internet-USENET Gateway at Stanford University
- Date: Fri, 4 Sep 1992 16:08:35 GMT
- Lines: 18
-
-
- During discussions on another topic, the question arose how remote execution
- mechanisms handle on-machine utilities that require authentication. In the
- context of Kerberos this question is formulated as follows. If a machine
- runs a Kerberized version of rlogin and rexec, is it possible to do away
- with user passwords altogether. In particular, once a user has been
- authenticated by Kerberos to, let's say, rlogin, how does he then use
- the su utility. Since it expects a password when changing the process
- effective uid to a different user, this seems to suggest that the user
- must also have a password on the machine. Furthermore, this password will
- travel in the clear from the machine to which the user's terminal is
- attached to the machine running rlogin.
-
- We tried to come up with a scenario whereby the su utility is kerberized, but
- kept running into the problem that su expects input to come from a
- terminal. Are we missing something obvious here?
-
- Dan Nessett
-