[tulip] LNE100TX rev. 5.0

[Jacob Schroeder]

On Fri, 14 Dec 2001 18:08:05 -0500 (EST)
Donald Becker <becker@scyld.com> wrote:

> On Fri, 14 Dec 2001, Jacob Schroeder wrote:
> 
> > Donald Becker <becker@scyld.com> wrote:
> > > On Fri, 14 Dec 2001, Jacob Schroeder wrote:
> > > > The hub is a repeater (a somewhat old 10Mbps Linksys with 8 RJ-45
> > > > ports and 1 BNC). ifconfig for this NIC currently shows 382,699 RX
> > > > packets, 993,522 TX packets and 232,851 collisions with 0 receive or
> > > > transmit errors.
> > > 
> > > These might be reasonable numbers for a busy network on a 10Mbps
> > > repeater.  Just check /proc/net/dev (not 'ifconfig' or 'netstat -i') to
> > > verify that you don't have errors.
> > /proc/net/dev reports no errors for eth0. 
> 
> > With 5 computers hooked up to the network but none talking, I can get
> > a little over 150 collisions just doing 10 pings to another Linux
> > machine. So that's why I was a little concerned, but from what I'm
> > hearing I guess that's not too abnormal.
> 
> That is abnormal.  The maximum non-error collision count is 15
> collisions per packet.  It's very unlikely that a packet will get to 15
> collisions, so if you are actually seeing 150 with 10 packets, something
> is wrong with the counts.
> 
> > I'm not sure I'll be able to upgrade to the 0.93 driver for a while, I
> > don't want to bother this list with my trivial problems of not being
> > able to successfully compile them. I've done the kernel on numerous
> > occasions, and read all the documentation for tulip, but I'm still not
> > successful.
> 
> Use the suggestions in
>   http://www.scyld.com/network/updates.html
> 
> I think that most possible problems have been addressed in the archives,
> but there are many possible ways for a distribution to mess around with
> the header files and paths.
> 
> Donald Becker				becker@scyld.com
> Scyld Computing Corporation		http://www.scyld.com
> 410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
> Annapolis MD 21403			410-990-9993
> 

Well, I'm been doing some working with it today, and I can't duplicate the problem. I must've left an ssh connection open or something when I saw those 150 collisions, in which case I should've paid better attention to the number of TX and RX packets. With only two computers talking today I can't get it to do more than an average of 1 collision per every 4 TX packets, and that doesn't count the number of RX packets either. Oh, well. I'm just glad that it doesn't seem to be causing any errors and my speed's not that bad for a 10Mbps network.

I'll let you know the results if I'm able to use the 0.93 driver. I'm still fighting with the driver in Debian trying to figure out why parts of the kernel source don't seem to be in the right place for the compile instructions to work (I've tried all the web pages on www.scyld.com and the compile instructions listed at the end of the source file).

Thanks again for your help and excellent work,
Jacob

-----
Windows:

32 bit extensions and a graphical shell for a 16 bit patch to an 8 bit
operating system originally coded for a 4 bit microprocessor, written
by a 2 bit company, that can't stand 1 bit of competition.

Linux:

The ultimate windows patch.

http://www.linux.org/