Duplex mismatch produces two different responses

[bms-integrationz]

Hi

Sorry if this is a simple question but I am currently working towards my CCNA for the first time and have recently noticed the following response from when testing duplex mismatching in a lab (note, the lab uses real equipment - this is NOT being simulated in Packet Tracer).

The response of the link produces two different results. I can see that one difference is that the speed of the links change from being both 100Mbps to both being Auto. So if this is the cause of the issue, I'd appreciate it if someone could please explain why?

Scenario 1: produces a UP/UP state and reports duplex mismatch (CDP-4-DUPLEX_MISMATCH) - I understand this response.

Cisco 2911 with settings as follows:

  (config-if)#duplex half

show interface g0/0:

GigabitEthernet0/0 is up, line protocol is up 
....
  Half Duplex, 100Mbps, media type is RJ45

Cisco 3750v2 with settings as follows:

   (config-if)#duplex full

show interface fa1/0/1:

FastEthernet1/0/1 is up, line protocol is up (connected) 
 ....
  Full-duplex, 100Mb/s, media type is 10/100BaseTX

Scenario 2: produces a DOWN/DOWN with no reporting of duplex mismatch. Link is reports as DOWN, line protocol is DOWN. This response is the one confusing me.

Cisco 2911 with settings as follows:

   (config-if)#duplex full

show interface g0/0:

GigabitEthernet0/0 is down, line protocol is down 
 .....
  Full Duplex, Auto Speed, media type is RJ45

Cisco 3750v2 with settings as follows:

   (config-if)#duplex half

show interface fa1/0/1:

FastEthernet1/0/1 is down, line protocol is down (notconnect) 
......
  Half-duplex, Auto-speed, media type is 10/100BaseTX

[balaji.bandi]

As you can see one side is gig interface - other side fastethernet.- you need to fine tune the negotiation to work as expected :

check below more explanation  and remediation :

https://community.cisco.com/t5/networking-knowledge-base/the-cisco-discovery-protocol-displays-the-quot-cdp-4-duplex/ta-p/3130830

[bms-integrationz]

Hi 

Thanks for the quick response.

However, the detail I am trying to understand is why does the duplex mismatch get notified when the gigabit interface is set to half-duplex and the fast ethernet is side is set to full-duplex, but it does not get notified in the log messages when the gigabit side is set to full-duplex and the fast ethernet is set to half duplex?

I would have expected the CDP duplex mismatch to be reported in the log messaging no matter which way around the duplex mismatch is set up for.

The one thing I can see is the when the gigabit side is half-duplex, and the fast ethernet side is full duplex, the link can still auto-negotiate the speed (to 100Mbps). But the link state remains UP/UP and the log messages continuously report CDP-4-Duplex mismatch.

But if the duplex settings are switched around the gigabit interface sets its speed to auto and the fast ethernet side does the same (speed is auto). So somehow the speeds being negotiated as auto on both sides causes the link state to be DOWN/DOWN and the log messages don't show CDP-4-Duplex mismatch.

So what is the root cause of the different behaviour on the link if you swap duplex settings around (and vice-versa)?

[Ramblin Tech · Accepted solution]

Typically, hard-coding the interface settings either disables auto-negotiation entirely or limits the capabilities exchanged during the negotiation process with each end of the link having independent behavior. Let's go with the idea of auto-nego being disabled on at least one end.

Speculating...

Scenario 1: The 2911 is hard-coded for HDX, but at what speed? I do not believe it actually supports 1000Mbps/HDX (many Cisco devices do not); that being the case, it could default to 100Mbps/HDX. The 3750 hard-coded for FDX could also default to 100Mbps, so the interface comes up, albeit with a duplex mismatch.

Scenario 2: The 2911 is hard-coded for FDX, but the speed question remains. Let's assume it defaults to 1000Mbps rather than negotiating down to 100Mbps. The 3750 cannot operate at 1000Mbps, so the link remains down/down and consequently no CDP to report a duplex mismatch.

If you changed Scenario 2 for the 2911 to hard-coding duplex full & speed 100 and the link came up, it would lend credence to the above.

[bms-integrationz]

Hi RT

Your suggestion seems spot on.

I retested both combinations and in the case of both interfaces defaulting to auto for their interface speed, forcing the gig interface to 100Mbps brought the link state to up/up and duplex mismatch messages started to be notified in the log.

So it seems that the default interface speed for the gigabit ethernet side is 1000Mbps (or probably more accurate to say its auto speed is not the same as the fastEthernet auto speed - so no auto negotiation occurs of speed happens when the gigabit side is at full duplex and the fastEth side is at half duplex) so the link just goes down.

Thanks.