NetXMS Support Forum

Hello All,
I've installed NetXMS couple days ago and I must say: Good job.

This is my first encounter with system that automatically detects network topology, and I have some questions. I have MikroTik network(about 200 AP), and couple of UBNT AP. I have one central cisco router serving PPPoE to clients over vlans(one vlan per AP interface), and serving couple management vlans to all other network devices. Big L2 network.
As switches i have Nexus 3k, Catalysts 3k, CRS's(switch chip used), rb2011(bridge), and rb750(bridge).
Is it possible for NetXMS to discover all this(i mean dependencies of each host)?
Now when I do some upgrading some devices i've got info about all nodes behind as down, and I cannot generate L2 map.
Before NeXMS I've used software where dependencies are defined statically, so I'm little confused.
Thanks in advance

Hi,

what you mean by "cannot generate L2 map"? If you create new map with type "layer 2" and seed node set to your central router, what you will get on that map?

Best regards,
Victor

Hello Sir,
Perhaps I misunderstood concept. I tried to generate a L2 map  for test from any random switch(rb750 -software bridged) in the network. Results in attachments. If I missed something sorry for trouble.

Its actually a MikroTik problem, they dont expose enough L2 topology information in SNMP.

You need at least one of CDP/LLDP/STP MIBs (or some other MIBs like BridgeMIB, etc.) exposed over SNMP for NetXMS to learn the L2 topology. CRS for example dont even expose the switch chip FDB over SNMP (they only expose the bridge and CPU FDB).
UBNT is even worse in this, their SNMP support is absolutelly horrible.

You Cisco switches and HP switches, etc. have these, so if you try to generate the L2 map from one of the serious switches (or the central router), it should be populated.

Thank You for answer Sir, after trying to pull l2 topology from Catalyst 2970 same result. Only one node shows on map.
How to deal with this devices then? I want to send SMS when node is down, but I dont want get messages on child nodes of that host(NODE_DOWN state). Event processing policy? What about performance when i have hundreds of hosts, and I use event processing policy for maintain host dependencies?

I would suggest fixing the issue with topology not being correctly learned rather than making workaround in EPP.

Is CDP/LLDP/STP enabled on that Cisco?
Is it properly exposed over SNMP?

How is the entire topology, does that Cisco even have some non-MikroTik neighbors (so that topology can be learned)?

Yes workaround is bad idea.
On all Cisco devices all of mentioned disabled on Mikrotik's /ip neig disc disabled on all devices, on UBNT CDP off.
I thought that FDB is enough to populate topology.
If MT would have LLDP ability someday....
All Cisco's are "surrounded" by MT or p2p equipment with limited snmp support.

When Ive upgraded to 2.07 and I saw couple of hosts on L2 map(map Ive created before) without changing anything in their config. All hosts have similiar configuration, and software version. Yesterday Ive upgraded NetXMS to newest BETA and all host dissapeared from map.
This post is only info(version 2.07 started to discover network). Im waiting for LLDP in MT(now in release candidate).
When Ive tried to pull L2 from Nexus Ive got info about problem retrieving FDB. My Nexus was discovered with driver Generic.