Any known problems - CDP between classic/old IOS switches and Meraki?

[TinyLittleAdmin]

Dear community,

in one of the networks I'm working on there is a mix of very old Cisco IOS Switches (3560a, 3750s, 2950s - 15y and older) and a new Meraki installation, where seven type MS250-48FP switches are stacked and do the core switch/routing for the nextwork. Some of the old IOS switches are chained ... so not all of them are directly connected to the Meraki. Also there is a NAC solution in place.

The problem is: within the Meraki dashboard AND in the NAC there are a lot of wrong or missing information regarding to CDP/LLDP. Which means: both show devices on switches, where those devices in fact never have been connected to. Even if I list all the classic switches in Meraki dashboard and enable the CDP/LLDP column - only two switches have infos in it.

Now I'm not sure.... am I dealing with firmware that's too old to deliver correct CDP? Those classic switches even don't support LLDP. If I'm correct CDP is always sent on VLAN1 - either tagged or untagged. Is it possible that my VLAN1 on the classic switches is in any way different to the VLAN1 on the Merakis? I also guess we are not talking about routing stuff as CDP is just Layer2.

Any ideas? After a few days of investigation, testing and searching I'd love to get suggestions or help.

Thanks in advance.

[Georg Pauwen]

Hello,

what if you 'clear cdp table' and 'clear cdp counters' on the older switches ?

[TinyLittleAdmin]

Hi,

I've just entered both commands on a "WS-C3560G-48PS-S" thats directly connected to Meraki on trunk - waiting now to see changes in Meraki after 180sec at least.

Are there any dependencies between CDP,ARP and DNS? I'm not an expert but it looks like maybe somewhere in the data exchange between old switches and meraki - mac addresses cant be resolved to correct IP or IP to hostname.

Update: no difference, still no info in "cdp" column within Meraki for that switch.

[TinyLittleAdmin]

Found another strange configuration in the same network. Not sure if that can cause incorrect CDP results.

1.) the meraki, which should also act as the STP/RTSP root ... indeed isn't. At least not for every vlan, especially vlan 1. In total I found 9 different root bridges on different switches where meraki isnt the root for the specific vlan

2.) at least three different native vlan's on trunk ports. i guess if another vlan is forced to become native vlan on a trunk port, than by default vlan 1 will be tagged on that port?

[TinyLittleAdmin]

Its' now almost six months - and just for the case that someone else needs the info: in my case it was a total mixup in STP and STP modes/protocols. All involved switches were set to use standard config, there was no real root config. And we've used pvst, pvst+, rtsp ...

SOLUTION:

we've installed two additional 3850s and connected them to meraki core. than we've disconnected all old catalyst switches (which do not support rtsp) from meraki, and moved them to those 3850s. there is now just one connection between the old and the new world. we've also set all old catalyst switches to use the same STP mode and reconfigured all switches to do a good STP. Problem seems to be solved. next step will be to replace old catalysts with meraki and switch their connection directly to the core meraki.