Thanks for the rapid response.
Here is a summary of the responses to my questions.
from:
Daniel S. Riley <dsr_at_lns598.lns.cornell.edu>
Jeff Stelzner <jeffs_at_esca.com>
Phil Antoine <antoine_at_RadOnc.Duke.EDU>
Clifford Krieger <ckrieger_at_psi.prc.com>
Hellebo Knut <Knut.Hellebo_at_nho.hydro.com>
> 1) What is the system cost of such high network interface error
> count? Can this impact system performance?
Yes, it could. NFS in particular is very sensitive to errors, but it can cause performance
degradation with other protocols as well.
> 2) Is this still a curiosity or a crises? At 200 Ierrs this seamed
> to be something to watch, but at 4400 should this now be a priority
> to fix?
I would make it a priority, especially if it recurs.
> 3) Do the users with these BAD Ethernet cards need to be attached to
> the DEC or simply logged on to our Novell 3.11 network?
Depending on the type of error, the cause could be anywhere in the same collision domain as
the DEC (bridges and routers bound collision domains, repeaters and hubs don't). It doesn't
have to be a misbehaving ethernet card -- it could be a bad cable or connector, or a transient
event associated with a cable being moved.
> 4) Anything you can suggest to help?
Do a 'netstat -sI ln0', changing 'ln0' to the name of the network interface with the high error
count. This will tell the particular type of the errors, which may help identifying the problem.
Received on Wed Aug 14 1996 - 23:01:10 NZST