SD-WAN latency on all TLOCs increases simultaneously

[brian.holmes]

Has anyone experienced the mean-loss on every TLOC increase simultaneously?  The volume of drops in the BE queue is also elevated at the same time.   Does anyone know if the BFD packets measuring the SLA get queued under best-effort/DSCP 0?   This impacts the AAR behavior obviously and I would never expect to see all tunnels unable to meet the SLA for drops at the same time over diverse circuits unless maybe the CPU on the router was running hot ( but I see no evidence in this case ).  17.12.4b

This problem is periodic throughout the day.

ROUTER#show sdwan app-route stats remote-system-ip x.x.x.x | i mean-loss|color
local-color             private2
remote-color            private2
 mean-loss   5.000
local-color             private2
remote-color            private1
 mean-loss   1.000
local-color             biz-internet
remote-color            biz-internet
 mean-loss   6.000
local-color             private1
remote-color            private2
 mean-loss   1.000
local-color             private1
remote-color            private1
 mean-loss   4.000

[brian.holmes]

Are SD-WAN BFD packets being placed in the best-effort queue by mistake ( bug )?  bad hardware?

[MHM Cisco World]

Is this issue solve

MHM

[Kanan Huseynli]

Hi,

control traffic is part of LLQ/ Queue0.

You can confirm by running packet-trace on transport interface. 

[wajidhassan]

If BFD SLA probes are affected across all TLOCs, it could point to a shared congestion point like an upstream shaper or misapplied QoS policy. Also confirm transport-side QoS isn’t downgrading control traffic unexpectedly—packet-trace can help verify queue assignment.

[akoukis]

Hello, you have enable simple AAR or E-AAR? Is an open bug for E-AAR and C8500.

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCwp01089

Best Regards

Anestis