GLC-LH-SMD

[harmonydave]

Hey everyone,

I have an odd situation here, 2 switches (3750X and 2960) connected with single mode fiber and using a GLC-LH-SMD SFP in each switch. When I check the DOM with 'show int gi0/1 trans det' (using the appropriate interface identifier for each switch), the 2960 end shows an optical transmit power of -5.1 and receive power of -20.0 (at the threshold for low warn). The 3750X end shows an optical transmit power of -5.5 and receive power of -5.0. At first I thought there was a fiber problem, but swapping the fibers didn't change the situation (and the fiber has been certified by the installers). I got an extra SFP and swapped it in, but that didn't make a significant difference either. Any ideas on how to troubleshoot this further, or possible causes?

thanks!

[supoonia]

Hi Dave,

lets start by giving physical self loop on each interface, i mean connect Tx with Rx, use some 2-3dB attenuator if possible.

and will able to isolate if it is interface or the fiber.

[harmonydave]

Hi Sunil,

Thanks for the quick reply. We don't routinely work with single mode, most of our stuff is short haul so I don't have an attenuator. I can try and beg/borrow/steal one from somewhere, but wouldn't you think that if it was the fiber that when I swap tx/rx at both ends the problem should move with it (ie the other switch should show a degraded rx signal)?

[supoonia]

yeah by swapping fibers at both ends, if it was fiber than it should have moved.

i think you can try without attenuaotrs.

-5dBm will be ok for Rx.

[Tom Randstrom]

What is the link length and the expected loss (per the documentation provided by installers)?  What makes up the link: how many splices and patch panels? 

As suggested, I would loop at the ends to make sure the SFPs are funtioning properly.  No attenuator is necessary on the

GLC-LH-SMD SFP. 

[harmonydave]

Thanks Tom et al, the link length is under 1000feet, I wasn't present when the link was installed so I'll have to review exactly what they did in terms of splices etc next time I'm on site. I don't have the documentation yet from the installers either, will get back to you on those points. I'll do the loop-back in the next day or two, once I'm on site and have dug up some SM fiber.

[harmonydave]

Update: tried the loop-back with both interfaces, and just to really make things interesting, there was no signal loss at either (Tx and Rx around the -5dBm to -7dBm). Put the installed fiber back in and the 2960 end again showed a degraded Rx close to -20dBm. I'm puzzled, any thoughts?

[supoonia]

Hi Dave,

than it looks it's because of some bad connector on fiber patch panel or some bad jumper somehwere,

you need to start with measuring power levels at each jumper/connector point and need to figure out.

all the best :-)

[harmonydave]

Thanks Sunil, sounds like a job for the installers . As far as I know, there are no breaks in the run anywhere except a splice box at either end (no patch panels, no other splices). I'm going to try the fiber swap test again, it must be a problem with one of the fibers, be nice to see it actually move with the fiber when I swap them!

[supoonia]

ok ..yeah something in fiber :-)

[Tom Randstrom]

If there are not any patch panels at the ends of the fiber, did they just splice a fiber pigtail with connector on to the end of the between building fibers within the splice box?  I was going to say maybe a bad jumper, depending upon how you moved the fiber jumper cable at the SFP ports (or patch panel ports), but you don't have any jumpers. 

Strange.