I`m testing v5 pre-RC2 and
seems that the PCQ does not take the real and current network bandwidth utilization, just divide the max-limit on the current value of the pcq-substream-count.
This leads to an average flow rate is lower than the pcq-rate even if the real bandwidth has not yet reached the value of max-limit.
Subscribers cut the rate even if we have not yet reached the value of max-limit !
After all, usually only a fraction of subscribers fully load their feed.
I would like to PCQ limits the speed of each flow (substream) by pcq-rate value, and began to proportionally reduce the rate of flow only after reaching max-limit or other limits of queue tree.
Is it possible?
Or how to implement such a scheme?