Path: utzoo!utgpu!watserv1!watmath!att!linac!pacific.mps.ohio-state.edu!zaphod.mps.ohio-state.edu!wuarchive!cec2!news
From: dale@wucs1.wustl.edu (Dale Frye)
Newsgroups: comp.protocols.nfs
Subject: Re: NFS for $'s or RFS for free
Message-ID: <1990Nov30.181848.15495@cec1.wustl.edu>
Date: 30 Nov 90 18:18:48 GMT
References: <1138@cnw01.storesys.coles.oz.au>
Organization: Washington University, St. Louis MO
Lines: 26

I am using RFS right now with SVR3.2.2. It works fine for sharing
files. Can also share devices. (with some limitations)

Problems I have with RFS.

I'm using it in a student lab (24 hour access) on 24 386 boxes. The
students treat the machines like DOS boxes i.e. if they get hung up
they reset the machine (UGHHHHH!!!!). RFS is not stateless. It maintains
a connection list on each machine advertising (exporting) a service.
RFS (over TCP/IP) does not detect the broken connection caused by
the reset on the remote machine. When the remote machine reboots and
remounts the resource it makes a new entry in the table. Enough of these
and the table fills. Machines can't then remount resources. 
  Because of this table you can only have x number of machines connected
to a server machine. (NOTE: Each mounted resource counts as a seperate
connection in this table i.e. if your table has 50 entries you can have
50 machines mount 1 resource or 10 machines mount 5 resources, etc.)

My response:
 If your machines are closely held ( the machines can't be rebooted
easily) and the number of mounts needed is not large RFS should work
OK. I'm switching my lab to NFS during winter break because of both
reasons.

Dale Frye
Washington University in St. Louis