I am running the packet_steering script /usr/libexec/network/packet-steering.uc
Output:
pid 596's current affinity list: 0
pid 596's new affinity list: 0
pid 597's current affinity list: 0
pid 597's new affinity list: 0
pid 1035's current affinity list: 3
pid 1035's new affinity list: 3
pid 1036's current affinity list: 3
pid 1036's new affinity list: 3
pid 1037's current affinity list: 3
pid 1037's new affinity list: 3
pid 1038's current affinity list: 3
pid 1038's new affinity list: 3
pid 1039's current affinity list: 3
pid 1039's new affinity list: 3
pid 1040's current affinity list: 3
pid 1040's new affinity list: 3
pid 1057's current affinity list: 3
pid 1057's new affinity list: 3
I have a MT6000 with NAPI threads for ethernet. Filogic 830 is a 4 core cpu and the cpu_siblings value is "f".
Shouldn't the affinity list be "f", not "3" or "0"?
When I download at 300 mbps with SQM one core is at 50%, two are at 25-30% and one core is at 5-9% load. Is the new script dynamic where if the load is high enough on one cpu then it bumps the process to another or changes affinity to more cpus?
