Re: [Bonding-devel] quick help with bonding?

[Jay Vosburgh <fubar@us.ibm.com>]

Chris Friesen <cfriesen@nortel.com> wrote:

>Jay Vosburgh wrote:
>
>> 	2.6.10 is pretty old, and there have been a number of fixes to
>> the bonding ARP monitor since then, so it may be that it is simply
>> misbehaving (presuming that you're running the 2.6.10 bonding driver).
>> Are you in a position to test against a more recent kernel (and/or
>> bonding driver)?  Does the miimon misbehave in a similar fashion?
>
>Testing a more recent kernel is problematic.  A new bonding driver could
>be possible, assuming the code hasn't changed too much.
>
>I just did another experiment.  Normally we boot via eth4 (which then
>becomes part of the bond  with eth5 at init time).  If I boot via eth6
>instead, it appears as though the problem doesn't show up.

	Well, if you're still inclined to investigate, you may want to
inspect the ARP probes generated by bonding in the "bad" situation.  I
don't really have any evidence to back it up, but one guess is that the
IP detection stuff in the ARP monitor is getting messed up. I'd check to
see if the ARP probes have the correct source IP address (which, in the
2.6.10 era bonding, is determined only once by inspection of outbound
ARP traffic, and never updated).  If you're not using active-backup mode
(you didn't say, and I can't tell from your log excerpt), then the ARP
monitor may not work at all (since it will send ARP probes with an IP
source of all zeros).

	If bad ARP probe source addresses are your problem, then that is
fixed in a later version of bonding, although the changes would require
some rework to backport to 2.6.10 (if they can be backported).

	-J

---
	-Jay Vosburgh, IBM Linux Technology Center, fubar@us.ibm.com