We are experiencing widespread, intermittent packet loss and network degradation on our "Broken Bow Urban Ring," a production network utilizing Adtran equipment. Our investigation has identified two primary contributing factors:

1. The presence of legacy hardware with limited processing capacity.

1. Excessive SNMP polling from our Zabbix monitoring system.

Evidence suggests that the high volume of SNMP requests from Zabbix is overwhelming the management planes of these devices, particularly the obsolete hardware, leading to control plane congestion that manifests as data plane packet loss.

 

1. Polling Rate Reduction: Is it possible to globally reduce the frequency of SNMP polling for the entire group of devices (or a specific host group) within this ring? We need to increase the polling interval to lessen the constant load on the devices.

1. Port/Interface Filtering: Currently, Zabbix may be discovering and monitoring every port on these Adtran devices. We would like to filter SNMP discovery and monitoring to only track specific, critical ports (e.g., uplinks). Is there a way to configure Zabbix to ignore a specific set of ports (for instance, all non-uplink customer-facing ports) to dramatically reduce the number of OIDs being polled per device?