[c-nsp] Switch Port Question

[Lincoln Dale]

On 25/01/2011, at 12:48 PM, Aaron Riemer wrote:
> This is a really basic question and I should know the answer. I am just
> pondering over loop guard and UDLD and take this for example:

note that at a fundamental level, mechanisms in Spanning Tree are always going to be different to those that work on physical ports.

for example, if you had a port-channel (LAG bundle) of 4 physical member interfaces and ran STP over that, how many of those interfaces in the bundle would be moving BPDUs?
answer: 1.

this should show quite clearly why mechanisms such as UDLD are important.

> Two switches are connected via a pair of fibre. Now let's say the TX path
> from switch1 (RX on switch2) fails or is disconnected but the other fibre is
> still operational. I would have thought that logically Switch 2 should see
> that it is no longer receiving "light" on its RX path and bring the port
> down? 

yes,
swltch 2 would consider the interface to be operationally down (loss of light).
switch 1 is still receiving light.

> 
> 
> 
> Obviously this is not the case.

it should be.


> Therefore my question is what determines if
> a port is physically "UP"? What physical checks are done by the switch port
> and how does it determine if media is attached?

the PHY/transceiver tells the switch that it has link.


cheers,

lincoln.