[j-nsp] Force ports with only RX connected to UP

[John Wilkes]

On Sun, Jul 11, 2010 at 12:28 AM, Phil Mayers <p.mayers at imperial.ac.uk> wrote:
> Your direct test I would expect to fail; pulling the TX fibre will cause the
> far end to signal fault and bring the entire link down.

Yes, the far end. But I meant I connected both TX and RX between two ports.
That made them both come up (of course). But when I pulled just one TX
they both went down.

> Are you sure the taps are working i.e. providing a valid light-level and so
> forth? What's the media & distances/powers?

It's LX and a less than 100 meters. Sure, it's split, but still. We
haven't measured it, but since
the unsplit port-to-port test doesn't work either then it's not that.

> What kind of tap is it? Passive fibre, OEO?

Passive.

> What does "show int ... extensive" show, with the tap inserted into RX?

It's not connected at the moment, but no change from when the incoming
light is off.
Just down and sad.

We've since tried connecting two ports with just one fiber strand, and
with Cisco and
Huawei it just works, once duplex is set to full (or just not auto).
The problem with that
is that Cisco doesn't (to my knowledge) have any cheap 24 port SFP switch, and
Huawei is Huawei.

We even checked the Huawei connected to the same tap, and it had no
problem there either.

> Are any odd protocols running at the *far* end? Can you turn them off?

Can't experiment with the live traffic. :-)
But yes, there is at least STP there. But again this would not affect
the two-ports-on-the-same-switch
scenario. It's not that it's up and blocked, it's down with all the lights off.