Quoting Malcolm Herbert (mjch@mjch.net):
> On Sun, Jul 07, 2002 at 06:37:44PM +0100, Patrick Welche wrote:
> |I don't want to discard rsh! I just don't understand why using ssh
> |instead should cause the two rdist processes from talking to each
> |other!
> 
> I'm using the rdist6 package and ssh quite happily. rdist6 will
> correctly use ssh when you supply the -P option[1]. it does tend to
> choke if the dialog with ssh is non-standard (eg, because you haven't
> collected a server key from the other machine yet) but otherwise it

My simple workaround was this:
mv /usr/bin/rsh  /usr/bin/rsh.orig
ln -s ssh /usr/bin/rsh


I used to do this all the time for krsh in a kerberized
environment.  Let rdist believe it's using rsh.


Basically, where rlogin/rsh was, put slogin/ssh.

I can use .shosts (or .rhosts) to log in with no auth
(and encrypt the transit).

I can use keys to allow for more secure authentication.
I can use ssh-agent to authenticate connections FOR me.

I stopped using rdist long ago as rsync meets 99.9% of
my needs, and does it better.  That other 0.1% was "run
a script remotely".  I can do that via rsh/ssh and tend
to just run a makefile or cfengine on the far end.

If /etc/inetd.conf is newer than /var/run/inetd.pid, then run
kill -HUP `cat /var/run/inetd.pid`

It's really not that hard.

Plusses: Rsync only copies over changes.  rsync can keep times
the same and will handle links well.  Perhaps rdist works better
now that last I used it, but I have no need for it anymore.