Gert Doering wrote:
Hi,
I'm facing a new challenge and hope that one of you has a nice trick for
me...
We are working on taking a 10G wavelength into service, and it has packet
errors (CRC errors showing up on one side). 3 different fiber providers
involved, plus local in-house cabling. Difficult.
For "classic" lines (T1 to STM-<whatever> serials), the normal approach to
diagnosing this is to loop back the fiber at certain points of the path,
and see which part is the problematic one.
This works with 10G fiber as well, but I can't find a way to ping-test
the loop from the Cisco.
Here's the setup:
Cisco 6506, Sup720-10GE =======<fiber path>======) loop cable
I can *see* the loop ("show cdp neighbor" shows myself, and if UDLD is
enabled, the interface goes to errdisable).
I just can't ping-test, as the router is clever(?) enough to locally
loop the packets - self-ping won't leave the box. On a serial interface,
pinging myself will make the router send out the packet over the serial,
so I can nicely ping-test the loop.
So - what's the trick for testing such things with a 10GE LAN interface?
WAG, but maybe something like:
int Te1/1
ip unnumbered Lo1
ip route 192.168.1.1 255.255.255.255 Te1/1
The un-numbered command ought to stop the router ARPing for next-hop,
and you can then:
ping 192.168.1.1
Obviously you won't get any replies because the cisco won't receive on
192.168.1.1 but the counters should increment.
I'd be interested to know if this works...
The other option might be to use two 10gig ports and a kind of Y-cable
arrangement for the patch lead, using a port for TX, and a port for RX,
and put either the TX or RX port into a vrf.
_______________________________________________
cisco-nsp mailing list [email protected]
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
