This is a follow up to my postings about poor NFS/FTP/PING performance.
We have done additional investigation and uncovered some new facts.
Let me start by describing our network setup:
=======================thick=net=================================
||
/------------\
thin wire-------| repeater |----thin-net----{3100}
| \------------/
| | |-----...
{RS6000-1} |-----...
|
{NCD x-terminal}
|
{NCD x-terminal}
|
{5000}
|
{NCD x-terminal}
|
{ALPHA}
|
{MAC}
|
|
{RS6000-2}
The original symptoms are that I had NFS trouble accessing large files
over NFS on a volume physically located on the 3100, while I had no problems
communicating with the 5000. I finally got hold of the layout of our net
layout above, and started various ping tests. The tests include some of
the nodes shown above, but also involve a news-server located in another
department (some bridges in between) and another RS6000 which is connected
indirectly to the thick-net shown.
the result is as follows:
originator target percent packet loss
----------------------------------------------------------------------
ALPHA 3100 1.4
ALPHA RS6000-1 0.0
ALPHA RS6000-2 0.0
ALPHA 5000 0.0
ALPHA news-server 0.9
ALPHA RS6000-3 0.8
3100 5000 0.0
3100 RS6000-1 0.3
3100 ALPHA 1.0
3100 RS6000-2 0.0
3100 news-server 0.3
3100 RS6000-3 0.5
5000 3100 0.0
5000 RS6000-1 0.0
5000 ALPHA 0.0
5000 RS6000-2 0.0
5000 news-server 0.1
5000 RS6000-3 0.2
The only consistent pattern is that any communication across the repeater
(it is NOT a bridge) involving my ALPHA encounters a packet loss rate
of about 1%, while communications within the same thin-wire involving my
ALPHA seems OK. There is no evidence that this problem occurs where other
nodes on this thin-wire are involved.
Does this mean I have a hardware problem on my ALPHA?
thanks
Charlie Jui
Dept. of Physics
U. of Utah
Received on Fri Feb 03 1995 - 19:43:51 NZDT