Hi guys,

I have a problem with Ganglia 3.1.2 since I upgraded the compute nodes from
Centos (x86_64) 5.3 to 5.5. All metrics seems to be correct but the
network_report.

I have two gigabit NICs in each compute node ETH0 and ETH1. ETH0 is
unconfigured and down. ETH1 is the interface I would like to see the metrics
for. I generated 25MB/sec network traffic on ETH1 which I can track with
"ifconfig" and "netstat -i".
Ganglia reports only 600k bytes/sec which is obviously wrong.

Any ideas if this is a Ganglia bug or a miss configuration on my side?

Please let me know if you need more information about my setup.

Thanks,
Jan

More information is always good :)  We can start with the model of your nic and
which kernel module it uses, etc.

Which kernel are you currently running?

I have the same issue with all compute nodes in the cluster. I have three types
of Broadcom NICs in the cluster:

NIC           kernel module
BCM95704A6    tg3
BCM5708       bnx2
BCM5709       bnx2

I am running on the 2.6.18-194.el5 kernel on all nodes.