Thanks for the reply, the problem was indeed that the switch's 2 ports =
were not configure to load-sharing (it's an Extreme Networks 7i switch). =
I am giving up on using mode=3D0 for this type of connection fow now, =
since it requires too much external support, mode=3D6 is easier to =
implement on a "normal" network.=20
I suppose that mode=3D0 works a bit faster than mode=3D6, is there any =
benchmark on the difference ? Do you guys have any idea what is the =
performance effects ?
> -----Original Message-----
> From: Tim Mattox [mailto:tmattox@...]
> Sent: Sunday, October 03, 2004 9:11 PM
> To: Shlomi Yaakobovich
> Cc: bonding-devel@...
> Subject: Re: [Bonding-devel] bonding and appletalk
> I think you have your network miswired/misconfigured for
> bonding mode 0. See below.
> On Sun, 3 Oct 2004 18:47:37 +0200, Shlomi Yaakobovich=20
> <shlomi@...> wrote:
> > When using mode=3D0, the sending machine receives its own
> > aarp probes, and it then decides that it cannot use this address,
> > and goes on to the next address, etc.
> For bonding mode 0 in regular ethernet networking, you need
> to either have the two (or more) NICs connected to separate
> and isolated networks such that broadcast packets sent on
> one NIC can't reach the other NIC on the same machine...=20
> or your multiple NICs need to be connected to a switch that
> understands and is configured for Linux Channel Bonding
> (sometimes this is called "trunking"). The latter method may
> not be sufficient for your needs.
> What you just described is the aarp packets from a node
> coming back to the node through one of it's other NICs in
> the bond. That is the problem. Suppose you have two
> NICs per machine in your setup, NIC A and NIC B. You
> would need to connect all NIC A's to a switch A, and then
> all NIC B's to a switch B.
> And then make sure that switch A and switch B do NOT have
> a connection to each other that would propagate broadcast
> packets. The easy way is to not have switch A and switch B
> have any connection to each other at all, except through
> the bonded machines in question.
> Good luck.
> Tim Mattox - tmattox@... - http://homepage.mac.com/tmattox/