Custom event handler script generates heavy CPU load with Bro 2.5.5 (PF_RING)

Hello. I have a script which defines a custom handler on mime_data event:

event mime_all_data (c: connection, length: count, data: string)

{

// do stuff

}

When this script is ran with capturing traffic in PF_RING mode using lb_procs=2, Bro processes consume 100% of both pinned CPU cores. This is not the case when capturing without PF_RING in single process mode though. What are possible reasons for this? Can it be optimized on the script side? What can be done to lower the CPU usage?

Hi Elena,

I think you're conflating two potential problems.

  - Is handling the mime_all_data event causing overload?
  - Is PF_Ring working correctly?

If you take your script out of the picture are you still seeing 100% cpu utilization?

   .Seth