[c-nsp] Possible causes for fiber link flap?

[Peter Rathlev]

On Thu, 2013-10-10 at 18:17 +0200, Peter Rathlev wrote:
> If the SFP+ to fiber interface counts as a connector we have six of
> those. Using Cisco's numbers[0] we should see around 2.8dB loss end to
> end, and their numbers are lower than yours.
> 
> I thus don't think 2.4dB loss in unreasonable; the DOM measured 0.6dB
> in the other direction was suspicious because of the asymmetry. After
> having moved things around some more we don't see that number anymore.
> It was probably an artifact of some kind.

Follow-up on this; the setup is this:

  +-------+    +------+    +------+   3 Km   +------+    +-------+
  | RTR-A |----| odf1 |----| odf2 |---====---| odf3 |----| RTR-B |
  +-------+    +------+    +------+          +------+    +-------+

The 2.4dB was from DOM on the transceivers on RTR-A and RTR-B.

The OTDR measurements made so far has been between ODFs 2 and 3 and they
probably didn't show 2.4dB. Unfortunately our own OTDR is misbehaving
and we haven't received data from the splice team (contractor) since
their last repair yet.

We have now introduced a switch between RTR-A and ODF 1 to rule out the
physical interface on RTR-A. This was necessary since we had no free
interfaces in RTR-A at the moment. It looks very much like the problems
are on the 3 Km run and we have asked the splice people to invesitigate.

Hopefully it's something simple like a clearly visible bend or such. :-)

And thank you all for the valuable input!

-- 
Peter