Thanks to erveryone who replied:
casper@holland.Sun.COM
koen@ciminko.be
hwe@uebemc.siemens.de
rob.e.allan@hydro.on.ca
glenn@atm.ch.cam.ac.uk
The answers can all be summed up by the following:
What you have to remember about spray is that the results are highly
dependent on the cpu speed and machine load of both the sending and receiving
machines. For instance, if you have a slow cpu machine (e.g. SPARC 1) it cannot
send the packages as fast as a sparc 10 machine. Also, the sparc 10 is fast
enough to receive them all. Hence if you spray a fast machine from a slow one
you will probably not see any packet loss. However, if you were to spray a
slow machine from a fast one, say spray the sparc 1 from the sparc 10, you will
see packet loss; because not only can the sparc 10 send out packets faster but
the sparc 1 cannot keep up with the rate at which the packets are arriving.
Add to this any delays that might be caused if you are spraying a busy machine
like a server.
All that spray really does is tell you a bit about the relative cpu speed of
the respective machines. Any packet losses are not differences in the spray
command or network problems.
Original Post
----- Begin Included Message -----
Can anyone tell me the difference between running the spray command
from a Sparc 1+ or Sparc 2 and an IBM Risc 6000? We get different
results when spraying the same host. We do not lose any packets from
the Sun systems. When spraying an IBM or Sparc from a Risc 6000 we loose
packets.
ibmhost 33% spray sunhost
sending 1162 packets of lnth 86 to sunhost ...
921 packets (79.260%) dropped by sunhost
23 packets/sec, 2017 bytes/sec
Note the packet lose. If I spray from a sun. I get 0% packet lose.
----- End Included Message -----
************************************************************************
Robert W. DiNuccio Micrion Corporation
Software Configuration Manager One Corporation Way
System/Network Administrator Peabody, MA 01960-7990
Email: bdinucci@micrion.com
Bus: (508)538-6778
Fax: (508)531-9648
************************************************************************
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:50 CDT