HFSC queue fairness after pfSense 2.6 to 2.7 netgraph change with mixed IPv4/IPv6
0 reputation · 29 Oct 2023, 04:56 UTC
A design review is needed to characterize traffic shaping behavior before applying optimization to a WAN link that carries mixed IPv4 and IPv6 traffic. The focus is HFSC scheduler interaction with saturated WAN bandwidth and the documented kernel netgraph changes between pfSense 2.6 and 2.7.
The relevant constraints are version-dependent queuing behavior and the presence of both IPv4 and IPv6 flows in the same limiter hierarchy. Uncertainty exists about whether low-priority queues can be starved under saturation and whether fairness assessment requires separate treatment of address families after the netgraph update.
What configuration boundary distinguishes acceptable latency from low-priority queue starvation for HFSC under saturated WAN conditions in pfSense 2.7 compared with 2.6? Is separate measurement of IPv4 and IPv6 queue lengths required to assess bottleneck before optimization? Does the service curve model need distinct parameters for mixed IPv4/IPv6 traffic after the netgraph change?
pfsense