From: Niall O Broin (niall at domain linux.ie)
Date: Thu 10 May 2001 - 12:29:11 IST
On Thu, May 10, 2001 at 04:49:54AM -0500, Stephen_Reilly at domain dell.com wrote:
> Thanks for the response Paul. However, does anybody know a method
> short of installing new things on the box of finding out the speed and
> duplex settings of a Network Card (RH5.2)?
I presume this is more or less a once off question i.e. you have a
particular box running RH5.2 and you want to find out how it is configured.
If this is the case (and I sincerely hope you don't have a bunch of such old
boxes to deal with :-) ) then the handiest way may be to plug it into a
switch which gives you that information - my little 50 quid toy switch from
Scan has 10/100, FDX/HDX and LINK/ACT indicator leds and I think the vast
majority of switches are so equipped. Certainly somewhere within Dell you
should be able to find such a switch. Of course if physical access to the
box or a suitable switch isn't on, then you're on a hiding to nothing.
You could always resort to the empirical method of course :
You should easily be able to determine the speed by sending something to or
from the box to another box, obviously on the same switch, and seeing if you
get a transfer rate of noticeably greater than 1MB/s, which is about wire
speed on a 10baseT network. If you're getting more than this, it must be
100baseT.
Full or half duplex is a little trickier though. You'll need to
simultaneously initiate two transfers, one A->B and one B->A, and check the
elapsed time until both are done. Divide the aggregate amount of data moved
by the elapsed time, and if the transfer rate is > 10MB/s => full duplex.
You can even combine both results to see if you're using 10baseT full
duplex. If you get a transfer rate of < 1MB/s in the first test , and > 1MB
in the second, you're most probably running 10baseT full duplex or
something's rotten in your 100baseT setup.
Now I don't think that the above testing methodology is going to get me an
invite to address the test and measurement engineers conference, so please
don't bother addressing all its obvious flaws, but it should be close enough
for government work.
Regards,
Niall
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:10:15 GMT