Path: utzoo!attcan!uunet!lll-winken!csd4.milw.wisc.edu!uxc!garcon!pequod.cso.uiuc.edu!dorner
From: dorner@pequod.cso.uiuc.edu (Steve Dorner)
Newsgroups: comp.sys.next
Subject: Re: Aaaaarrrrgggghhhhh!!!!!?
Message-ID: <1115@garcon.cso.uiuc.edu>
Date: 25 May 89 12:25:14 GMT
References: <10806@orstcs.CS.ORST.EDU>
Sender: news@garcon.cso.uiuc.edu
Reply-To: dorner@pequod.cso.uiuc.edu (Steve Dorner)
Organization: University of Illinois at Urbana-Champaign
Lines: 32

In article <10806@orstcs.CS.ORST.EDU> grahamr@prism.cs.orst.edu (Robert Graham) writes:
>NetManager never seems to get the NetMask right. Its fine right now, but
>when I do the local configuration, it tells me that it is the same as
>my IP number.

That's ok; the IPNETMASK number never gets used anyway, unless you hack
rc.boot.  The netmask gets set by black magic.

Actually, I suspect that the netmask is discovered by broadcast.  Evidently
if you play your cards right, a broadcast containing your IP address is
supposed to be returned to you with your subnet mask.  (I'm fuzzy on the
details; I'm just repeating what I was told by a local TCP/IP implementor.)

Of course, this assumes there's somebody listening to your broadcasts.  We
have a few "gateways" on this campus that are really stupid; they do nothing
but route packets.  These tend to be used in low-traffic areas, with
other stupid machines like PC's.  We put a NeXT on a net, and it was
unable to discover its netmask.  Eventually, we hacked rc.boot to make
it use the IPNETMASK value (at which point NetManager's failure to
set it properly becomes interesting again...).

There is also a problem with static routing.

I've posted patches to the rc* files in the past for dealing with this
problem, so I won't bother again.  If any of this sounds familiar to any-
one who missed the patches, let me know and I'll send you another copy.

Steve
-- 
Steve Dorner, U of Illinois Computing Services Office
Internet: s-dorner@uiuc.edu  UUCP: {convex,uunet}!uiucuxc!dorner
IfUMust:  (217) 244-1765