01-21-2016 06:45 PM
- We only talking about L2 switches
- Model include: (there are other switches, all varies in version and model; these are just some examples)
* ICX6450-48 POE Version 08.0.20T311
* FastIron WS 648 G-48, Version 07.0.01cT7
* FCX648G, Version 07.0.01T7f5
* FastIron 648G-48, Version 05.0.00T7e5
- Switches have CDP and FDP running. All can see the next hop neighbor or even a core connection.
- All Layer 2 and 3 devices are discovered. A few hubs, we don't care about.
- SNMP is working good. I can pull a full backup with BNA and push commands to it via BNA as well; no issues here. Aditionally, sflow also enabled on each interface and globaly.
Most of the addresses are all in 172.16.0.0 /16 range. There are a few VLAN's in 192.168.18.0/24
I'm attaching a small pic or the above examples.
I appreciate the help.
Couple of things:
* These hubs, do any of them interconnect L2/L3 devices? Or are they only at the end of the chain? i.e. one uplink to a proper switch, with only desktops plugged into the rest of the ports.
* Focus on a single device like 172.16.1.111. If you login to that device, what CDP/LLDP/FDP neighbors does it report? Login to those neighboring devices. Run the same commands. Do the two devices match up?
01-22-2016 06:44 AM
* The hubs are just end point devices; users trying to connect a laptop and PC, per say.
* 172.16.1.111 sees FDP neighbor -- 172.16.1.113 which also sees 172.16.1.111. Devices match up.
I when little further and traced the FDP descovery and compared it to the BNA link. (see attached .jpg)
So, the .111 discovers and also discoverable by .113
the .113 disovers and also discoveragle by .118
the .118 disocvers and also discoverable by .113
the .118 links to the core, as you see in the picture. but above devices do not inner connect.
01-24-2016 04:12 PM
I wonder if there's something going on with the inter-play between the different switch/OS families, including what looks like some fairly ancient code versions? Topology discovery will probably work with those code versions, but they don't test the older versions - e.g. BNA 12.4.2 is tested with FES 4.1 code trains.
For the other situations where you have missing links, are they all chains of the same device, or are they mixed groups like the 111/113/118 one described above?
01-25-2016 06:43 AM
Starting to sound like a ticket with Brocade. This is indeed a strange situation, especially when so many different codes and firmware.
The example I've provided is just a situational layout that happens to be chained. The other switches, like a single one connecting directly to the core, does not link either. The switch firmware is 08.0.20bT311, and it's also discoverable by both ends.
I'm kind of stuck at this point. Many points point to many different things. Might have to break it down in an individual instance. Unless, any other options are out there that I don't see.
01-25-2016 09:41 AM