[Linux-cluster] Rx error on broadcast ping

[Alexander Spanke <aspanke hpce nec com>]

Dear all,

I was got a reminder by a friedn about our network situation; we are
maintaining a cluster of ~200 nodes, connected via Gigbit ethernet. The
network configuration of the cluster is usual, nothing special; but if I
do a ping -b <bcast address> I see the Rx error counter increasing.

During normal network activity like NFS, scp or other applications,
everything is fine, only with doing broadcast pings I can reproduce the
increasing Rx error.

Does someone of you saw such a problem before ?

Thnx for any idea,

Cheers
 Alex