From Paragon Insights 4.1.0: https://www.juniper.net/documentation/en_US/healthbot/information-products/pathway-pages/api-ref/paragon-insights-api-4.1.0.html#devicegroup_schema_publish 
devicegroup_schema_publish -Up
destination
array[String]
field (optional)
array[String] format: string
sensor (optional)
array[String] format: string

The sensor information was removed, but the test is still showing the duplicate sensor information. Sensors were initially set, removed from config, still getting notification. The service level handling snmp trap has the lingering configuration because the config does not get synced from config server to the pod for changes in publish sensor config stanza.

Two events seen for the same log on the kafka consumer when sending notifications from the HB Kafka publisher.

One is the sensor and the other is the parsed one.

1) Get the specific Kubernetes pod name which matches grep with string "snmp" and "juniper-group"

kubectl get pods -A | grep "snmp" | grep "juniper-group"

2) Exec to the above pod and sync latest config-

kubectl exec <pod-name> -it -n healthbot -- bash

curl -X POST http://$NODE_IP:7005/sync -H 'Content-Type: application/json' -d '{"channel": "device-group-juniper-group"}'

/jfit_scripts/jfit_reconfigure.sh