3x100Mbps bonding: works!

beowulf at seinfeld.engr.ucf.edu beowulf at seinfeld.engr.ucf.edu
Thu Oct 5 10:56:46 PDT 2000

On Wed, 4 Oct 2000, Laurent Itti wrote:

> Hi all:
> ok, what was wrong is that the MAC addresses did not truly get all set to
> the same value for the three branches in the bonded channel.
> A peculiarity of the Realtek driver I guess.
> however, there is a very cool program "rtl8139-diag" written by Donald
> Becker (available on his driver page) that allows for low-level access to
> the realtek eeprom; it seems that the program was all setup to change the
> HWADDR except that there was no implementation for doing the actual change
> (i.e., the option existed, the argument would get parsed, and then nothing
> would be done with it!) That was just a matter of adding a few lines of
> code, and it works flawlessly.
> fantastic!

Have you checked the performace of the three channel network yet.  When we
tried using three channels we did not get as good of performace as the two
channel setup, so that is what we have been using.

I also had problems with the setting of the MAC on the first ethernet card
enslaved (KNE100TX).  The problem was that ifenslave was reading the MAC
from the first card and setting the bond's MAC with that.  It would then
read the MAC of bond back and set the MAC of the ethernet card.  On my
system when it it would read the MAC back from the bond it would come back
with 00:00:00:00:00:00 and set the first ether to that.  To fix it I put
in a delay in ifensalve between the setting of the bond MAC and the
reading it back.

I did not check the interrupt settings back when I did the tests, but now
all ethernet cards are assigend different interrupts, so I am not sure if
that was a problem when I did the tests.

Judd Tracy
jtr at isl.engr.ucf.edu

More information about the Beowulf mailing list