home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.kerberos
- Path: sparky!uunet!stanford.edu!CYGNUS.COM!eichin
- From: eichin@CYGNUS.COM ("Mark W. Eichin")
- Subject: Re: my previous question refined
- Message-ID: <9209050013.AA28302@tweedledumber.cygnus.com>
- Sender: news@shelby.stanford.edu (USENET News System)
- Organization: Internet-USENET Gateway at Stanford University
- Date: Sat, 5 Sep 1992 00:13:02 GMT
- Lines: 44
-
- Mr. Nesset makes several comments...
- > the remote machine. To protect it, you probably want to logon originally
- > by executing "rlogin -x". Of course, this means either knowing ahead of
- > time you want to change to root, or (more likely) always executing "rlogin -x".
- I do work cross-country (via AlterNET and NEARnet); the 56Kbps
- link at the other end is far more of a bottleneck than DES encryption
- will ever be. I always use encrypted rlogin.
- Note that the 91.03.25 version of telnet does support toggling
- encryption on and off, inbound and outbound, if you do notice the
- performance.
-
- > Jon Rochlis suggests either using ksu or "rlogin -l root". The latter option
- > doesn't work for Telnet.
- Actually, when the aforementioned telnet is invoked as rlogin,
- it *does* work. (It seems not to work for root, but there is no
- problem switching to other users based on the .klogin file; failing as
- root in particular appears to be simply a bug.)
-
- > access tickets) and then logon to the remote machine. I'm not sure, but
- > I think this means that a user with root privilege has those privileges
- > on all machines.
- The only machines that a user with a root instance has root
- access to (via rlogin or ksu) are those with their root instance
- listed in the /.klogin file. This could easily be no machines at all;
- root access has nothing to with the name of the instance, all that
- matters is that the name of the principal given is on the access
- control list.
-
- > Barry Jaspan points out that support of ksu could provide a security hole
- > if the machine doesn't have a rcmd.<hostname> srvtab. That's another concern,
- If the machine is "tightly kerberized" as you describe, then
- without such a srvtab, you won't be able to get in at all, other than
- from a directly wired terminal, so ksu won't be a threat. It is still
- important to be careful about making ksu available on "client only"
- machines that don't have such a srvtab.
-
- > do root work, you logoff the remote machine, su to root (which gets root
- > access tickets) and then logon to the remote machine. I'm not sure, but
- I don't think Jeff implied that it was necessary to log off;
- most of us use window systems of some sort, so popping up an
- additional connection isn't a hassle.
- _Mark_ <eichin@cygnus.com>
- Cygnus Support --
- "Cygnus Network Security"
-