If there is an SNMP problem will it at least show up as unreachable?

if you have configured at least one of the Layer 3 disco modules I would say it should show up...

but I am not sure...

but there are other things I think that are important:

1) are the seeds globally configured or per module

2) if per module, have you set a hop count ?

if there is a hop count defined, is it for all or only a few of the seeds?

4) do you have checked 'use DCR as seed' (which is enabled by default) ?

these are the general points you can configure for CSDiscovery - but there is no explanation/documentation available what they mean in detail and how they work!

In former releases of LMS seeds where only used for the initial discovery - I think this did not changed; but what happens afterwards?

The new feature 'Hop count' seems to restrict the subsequent discovery to the initial seeds and the sourrounding devices, being a Hop count of x hops away from a seed. So what means the option 'use DCR as seed' which can be choosen along with 'Hop Count' ? For me, this seems to be an opposed configuration option...

Currently, I would say if a 'Hop Count' is defined, new devices will only be discovered if they are not more than x Hop Counts away from any device listed as a seed.

Does the 'Discovered Devices' show the neighbors of each reachable device in the list or is this column empty ?

The device will show up as unreachable in Discovery IF it was found by one of the modules. It will not, however, be added to DCR.

I finally got a class and the instructor turned off CDP on three switches in his lab. Using all possible discovery methods we found two out of three switches. These were old catalyst 1900s. The two that were found had the same manufacturer number in the MAC address. The one that wasn't found had a different one.

I think all of the switches had SNMP set up, so I am not sure what happens when there is an SNMP problem.

Based on that experiment and what I have seen in my production network it appears that discovery detects the devices except for certain very old devices.

In order to determine what is going on, a sniffer trace will be needed. Chances are there is an issue with either missing objects, or slow SNMP response time.

What exactly is meant by Unreachable in discovery? Does it only mean that it is SNMP unreachable?

Well, it could be IP unreachable from the LMS server, but it could just be unreachable via SNMP.