broadcom network drivers and arps

Mark Watts m.watts at
Wed Jun 25 04:58:01 CDT 2003

Hash: SHA1

IPVS failover using heartbeat/ldirectord works flawlessly here with Mandrake 
9.1 and bonded broadcom cards (on a 2650). Bonding is done in active/passive 
mode with the 2.4.21 bonding driver,. (creating a 'bond0' on each box)

I have two 2650's failing over a single IP address. eth0 from each server is 
in one cisco switch, eth1 on each server is on the other switch and both 
switches are connected into one logical switch.

As an aside, I also have dual port Intel server cards bonded together (with 
the kernel driver) to act as a secondary interface (again cross connected 
into cisco 2950/24 switches.)


> Configuration:
> PowerEdge 6650
> 2 x Broadcom cards
> Red Hat AS 2.1 (kernel 2.4.9-e.25), HA cluster software
> Network drivers and software for fail-over installed from source
> packages available from (basplnx-6.0.1-1.src.i386.rpm,
> bcm5700-6.0.5-1.src.rpm)
> There are two network interfaces eth0 (active) and eth1 (stand-by)
> which are bundled to virtual sw0 (fail-over).
> Problem:
> When a cluster service moves to the second node (virtual IP - VIP
> goes with a service), network interface eth0 on the first node still
> sends arp requests claiming it has VIP, such requests don't occur on sw0.
> Problem disappears when i restart basp service on the first node
> (after relocation to node 2).
> Arek Wojtowicz
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at
> Please read the FAQ at or search the list
> archives at

- -- 
Mark Watts
Senior Systems Engineer
QinetiQ TIM
St Andrews Road, Malvern
GPG Public Key ID: 455420ED

Version: GnuPG v1.2.1 (GNU/Linux)


More information about the Linux-PowerEdge mailing list