Per-unit-scheduler and interface drops

  • 1.  Per-unit-scheduler and interface drops

    Posted 11-20-2020 09:58
    I have an interface with per-unit-scheduler and a lot of separate units with separate queues, on a MX480 with MPC3E-NG-PQ. The per-unit-queues work amazingly well.

    However, there are now enough subinterfaces that they could in theory hit the 10Gbps-limit of the physical port. I would like to monitor this, and the obvious choice is something like:

    admin@> show interfaces queue xe-1/0/5 aggregate | grep RED
    RED-dropped packets : 659675591 160 pps
    RED-dropped bytes : 984749892761 1890432 bps

    Unfortunately it appears that the statistics for the port queue includes the drops for every unit queue. This is not very helpful for determining whether I have reached the 10Gbps limit.

    Is there a way to see only the drops from the output queue on the physical interface? Preferably without having to iterate through all the subinterfaces to subtract their drops...