Customer saw constant packet loss on one of our ACX710 access routers despite only ~20% utilisation of that interface.

Interface queue RED drop can be seen from the router.

Physical interface: xe-0/0/7, Enabled, Physical link is Up

 Interface index: 141, SNMP ifIndex: 515

 Description: MCY-HOL | UFF000003089691 | TFF | NU 10G UFB | Q583 F5/6 | ae0

Forwarding classes: 8 supported, 4 in use

Egress queues: 8 supported, 4 in use

Queue: 0, Forwarding classes: best-effort

 Queued:

  Packets       :     2243914350957        238195 pps

  Bytes        :   3057701295993420      2540208672 bps

 Transmitted:

  Packets       :     2233253168012        237933 pps

  Bytes        :   3041845497834411      2537326944 bps

  Tail-dropped packets : Not Available  

  RED-dropped packets :      10661182945          262 pps

  RED-dropped bytes  :    15855798159009        2881728 bps

 Queue Buffer Usage:

  Reserved buffer   :       439 pkts        112384 bytes

  Shared buffer    :        1 pkts         1552 bytes

On Junos ACX710 when scheduler parameters (transmit-rate, buffer-size, and priority) are changed with traffic flowing, it may result in continuous drops for the VOQs (Virtual output queue) associated with the interface and the CoS profile that was changed when the traffic is way lower than the port bandwidth. All the VOQs associated with the CoS profile (scheduler) will be affected.

The 'show interfaces queue egress <interfaces>' command can be used to check the RED (Random Early Detection) drops.

This problem had been fixed vir PR1704589 in junos:21.2R3-S6, junos:22.3R3, junos:22.4R3, junos:23.1R2, junos:23.2R2, junos:23.3R1 and onwards.