Hi all,
We are using PF_ring to capture packets with 8 NIC in an redhat linux server, with 20 process for capturing and dumping packets. And PF_ring is in standard mode.
When we capture NIC one by one, then no packets is lossed, but if we capture all NIC concurrently, then one of the NIC will drop/discard packets and other NICs are working well, while the CPU usage is quite low. The traffic flow speed of the loss packet NIC will reduce from 600Mbps to 300Mbps.
Could you tell why this happen?
We are using PF_ring to capture packets with 8 NIC in an redhat linux server, with 20 process for capturing and dumping packets. And PF_ring is in standard mode.
When we capture NIC one by one, then no packets is lossed, but if we capture all NIC concurrently, then one of the NIC will drop/discard packets and other NICs are working well, while the CPU usage is quite low. The traffic flow speed of the loss packet NIC will reduce from 600Mbps to 300Mbps.
Could you tell why this happen?