There can be hardware and software problems.
Hardware:
. these could be due to faulty transceivers on the faulty hosts.
. but I have no tool, except permutation, to test this!
. uunet!mcc.com!vasey pointed me to an H-P Lanalyzer.
. check that you have no products designed for Level 1 ethernet.
. only DEC seems to have a book on hardware specifications for
mixing think and thin ethernets, multi-port transceivers, etc.
Software:
smb@ulysses.att.com pointed me to synchronized network accesses.
For example, rwhod on diskless workstations is very bad!!
And more precisely, broadcasts which triggers NFS activity is bad!
So I got a source for rwhod. And I added a flag (-n) to tell
an rwhod not to update files in /var/spool/rwho. On diskless
workstations, I set this flag and I mount /var/spool/rwho of
the server. So I have the same functions; and the collision
rate is better!
Thanks a lot to : smb@ulysses.att.com
vasey@mcc.com
And also to: northrup@cps.msu.edu
david@srv.PacBell.COM
wde@water.agen.ufl.edu
--Jacques Beigbeder
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:12 CDT