We have a device in the network that has multiple network addresses configured. Spectrum is creating multiple models for this device. One for each ip address configured on the device.
Release : Any
Component : Spectrum Discovery & Modeling
A check of the ipAddrTable and ipAddressTable tables in the ip mib of the device using MIB Tools shows both of these tables are empty.
When Spectrum models a device, it reads the ip addresses from both of these tables and updates the deviceIPAddressList attribute id 0x12a53 with all of the ip addresses configured on the device as seen in these two tables.
The deviceIPAddressListPollInt attribute 0x12298 rereads these tables at the time configured (once per hour) to update the table with configured ip addresses.
When modeling a new device, Spectrum checks the deviceIPAddressList attribute id 0x12a53 on the existing models to see if that ip address is already configured on a model. If so, it will not model the device. If not, it will model the device.
The root cause for these duplicate models in this instance is the device does not populate the ipAddrTable and ipAddressTable tables in the ip mib on the device.
This is an issue with the device and needs to be addressed by the vendor.
You can configure Spectrum to exclude modeling an ip address. For example, if the two ip addresses involved were 1.2.3.4 and 4.3.2.1, you can add one of the ip addresses to the "IP Exclusion List" field in the AutoDiscovery Control -> Modeling and Protocol Options subview of the VNM model. Reference the "Modeling and Protocol Options Subview" section of the documentation for more information.