We have a device which we have modeled in Spectrum, but only some interface models for the device were created. There are 40 or so interfaces in the interface table in SNMP, but only 3 show up in Spectrum. There is no interface filtering as far as I see (the FilterInterfaceRanges attribute is blank). When reviewing a sniffer trace I can see there are 2 - two - report packets being sent to Spectrum.
For v3, Spectrum sends a get-request to the device with blank engineid, engineboots and enginetime. The device needs to respond back with a report packet showing unknownengineid and in it, provide the engineid, engineboots and enginetime. This device responds back with engineid, but not actual engineboots or enginetime. It reports they are 0.
Spectrum then sends a get-request for sysobject id and uses the engineid and engineboots and enginetime of 0. The device then responds back with a report packet with the correct engineboots and enginetime and a "not in time" error.
Release : 10.3
Component : Spectrum Core / SpectroSERVER
Agent issue
Spectrum functionality needs the the initial report packet to be unknownengineid and in it, provide the engineid, engineboots and enginetime. This device responds back with engineid, but not actual engineboots or enginetime. It says they are 0.
Spectrum then sends a get-request for sysobject id and uses the correct engineid and sets engineboots and enginetime of 0. The device then responds back with a report packet with the correct engineboots and enginetime and a "not in time" error since the engineboots and enginetime are not really 0. This throws off the Spectrum code, and causes the problems with the modeling.
This needs to be resolved by the vendor - or model the device using SNMPv1 or SNMPv2.