Hey,
I currently have grafana running with node.exporter polling data to Prometheus.
The issue that I am seeing is that the bandwidth is much higher that is being reported to me via node.exporter. When i am running iftop i am getting a much lower number. The difference is very high. For example grafana was reporting the bandwidth at 450 and when I run iftop the bandwidth is reporting around 50.
Is your problem with Grafana or node_exporter? If node_exporter, then I suggest asking the Prometheus folks about it.
I believe it would be the prometheus end. I was asking on here incase someone ran into the same issues.