Path: utzoo!utgpu!news-server.csri.toronto.edu!utcs.toronto.edu!cks
Newsgroups: comp.unix.ultrix
From: cks@hawkwind.utcs.toronto.edu (Chris Siebenmann)
Subject: Ultrix 4.0 RIS looses
Message-ID: <1991Apr17.212012.11522@jarvis.csri.toronto.edu>
Organization: Ziebmef home away from home
Date: 18 Apr 91 01:20:12 GMT
Lines: 30

 So, I decided that instead of hulking the SCSI tape drive around to two
remote sites to load RISC Fortran 2.0, I should investigate doing a
remote installation from the tape's current location, my office
workstation. Setld has an option for installing from a remote host, but
it uses RIS instead of rmt or rsh + dd or something nice. Win some,
loose some. So I set up RIS on my office workstation and started
fiddling.

 Loss 1: RIS insists you can only register machines in the same domain
as your host. This is silly, since RIS can be used by anyone who can
reach your machine via TCP/IP; getting around this requires installing
bogus /etc/hosts names, installing the machines, and then fixing
~ris/.rhosts.  And why does RIS insist on knowing the Ethernet
addresses all the time?  I'm never going to try booting these machines
via RIS, so I could care less; I made up bogus Ethernet addresses for
them.

 Loss 2: A RIS-based setld load seems to loose dependancy information;
the remote install tried to install, in order, the Fortran manpages,
the Fortran programming environment, the F77 unsuported utilities, and
the V2.0 (Fortran) backend. The middle two depend on the last one, and
refused to install; I had to rerun setld afterwards and tell it to
install those two again. When I installed it on my workstation from a
local tape drive, setld got it right.

--
	"NFS should be viewed as a superior replacement for FTP, not
	 as a real network-wide Unix filesystem."
		- Henry Spencer
cks@hawkwind.utcs.toronto.edu	           ...!{utgpu,utzoo,watmath}!utgpu!cks