Validation of FCS

Mailman List Mirror (Read Only)
1

Hi all,

I'm trying to confirm (or debunk) my current understanding of FCS errors. An FCS error is a layer 2 error. In Ethernet spake, the 4 bytes of FCS data within each Ethernet frame is validated by a CRC check, which is done by the device receiving said frame. If the CRC check fails, an FCS error is reported by that receiving device.

If that understanding is true and presuming a "circuit" was made up of many layer 2 devices between the A and Z side of said circuit, it would be impossible for a CRC error somewhere along the path of that circuit to register on the receiving device of either the A or Z side. Perhaps in simpler terms, a CRC error is a localized thing and would never be forwarded from one device to another.

Is that fair and/or accurate?

Thanks in advance.

2

It would be forwarded in cut-through switching.

3

... until the bad frame reached the first store-and-forward switch (or most any router) which would log the FCS error, correct?

4

... until the bad frame reached the first store-and-forward switch (or most any router) which would log the FCS error, correct?

Log and drop yes. cut-through would log it also, but it would be too late
to drop it.

5

I have cut-through switches (arista) that log these as TX errors, they have already left the barn at that point.

6

it's a real pain when this happens because you have no idea of the ingress
port and consequently the source of the data corruption source. The same
happens for regular packet drops on some cut-thru switches (e.g. brocade
ti-24x). Annoying.

Nick