Having looked at the source for bond_activebackup_arp_mon in
drivers/net/bonding/bond_main.c of linux-2.4.31 & linux-2.4.26
and the documentation for the bonding driver provided on your site
There is a notable discrepancy here. The documentation explicitly states
" The ARP monitor could be configured with a target at the remote end of port3,
thus detecting that failure without switch support. " (referring to the
diagram in '12.2 High Availability in a Multiple Switch Topology')
However in the code an interface only becomes 'down' if it has no tx/rx for
longer than 2*delta_in_ticks, however if you have two such servers attached to
switches in 12.2, the arp packets generated by the arp monitor will server to
satisfy the tx condition, and the arp packets generated by the second server's
arp monitor will fulfill the first's rx condition. Thus the primary inter face
will always be up even if the link upstream (port 3 in this case) of the server
Is there a sensible fix to this available or will i have to add another feature
to the bonding driver to be able to properly detect upstream failiure?
Get latest updates about Open Source Projects, Conferences and News.