ERS 5510-24T shows the 1000SX GBIC in system information but the IN USE LED never lights
Joining two buildings over a Nortel pair that has been in place longer than I have. The plan was simply to light the existing house fibre between the two switches and be done with it, but port 23 will not come up on either end.
- Two Nortel ERS 5510-24T, software 4.0.2.02
- 1000SX SFP GBICs, one in port 23 on each switch
- roughly 120 m of building fibre through the risers, terminted in the far cabinet by whoever did the original job
port 23: no link, IN USE LED dark on both switches
GBIC listed in system information on both endsWhat I tried:
- turned auto-negotiation off on port 23 at both ends
- set symmetric flow control on the port
- built an MLT around it, in case the port had to be a member of something
None of it changed a thing. The manual says the GBIC has to be activated in software but never says where that setting lives, and I cannot find anything resembling it in the menus. Is there really an activation step I am missing, or am I chasing the wrong thing entirely?
Also available in: English
العربية
Deutsch
Español
Français
हिन्दी
Bahasa Indonesia
Italiano
日本語
한국어
Nederlands
Polski
Português
Русский
Türkçe
Tiếng Việt
简体中文
Comments 4
The IN USE LED does not report that a module is seated. It stays dark until the port actually sees an optical link, so all it is telling you is what the port status already told you - nothing is linking - and there is no hidden software activation step to go looking for. The manual wording sends everybody down that hole.
Two things really cause this, and I have met both in the same building.
First, the module has to suit the glass. 1000SX is a multimode part. If the risers are singlemode you can configure whatever you like and it will never link; what you need there is an LX module, AA1419015 on this platform.
Second, and the one that usually wins on old house cabling: transmit has to land on receive. A pair that was landed backwards in the remote cabinet gives you exactly this symptom - both GBICs visible in system information, both ends configured, no link at all. In the case I dealt with, swapping the pairs over in the far cabinet was the whole fix.
Order I would work in: find out what fibre is really in the risers, then walk the pairs end to end for direction, then borrow a run you know is good before you start blaming the module. Put auto-negotiation and flow control back to their defaults while you do it, neither of them is your problem here.
Before you dig any deeper in the menus: what is actually in the risers, multimode or singlemode? Nobody labels this properly, and 120 m between buildings is exactly the distance where people fnid out the hard way.
Second question: has anyone opened the far cabinet and verified the terminations, or are you trusting the labels there?
A test that costs nothing while you wait for an answer to either: take a short patch cord and loop one fibre port to another on the same switch. If that link comes up, the modules and the ports are fine and everything you are looking for is out in the plant.
Same shape of problem on completely different kit: three LANCOM GS-2326P+ switches linked over multimode through floor patch panels. Modules detected, light on the fibres, every fibre port stuck at no link.
Direction is what everything pointed at in the end. Seeing light on a strand tells you nothing about which way it is travelling, and a patch panel in the middle of a run is exactly where a pair gets crossd over, so walk every panel and prove that transmit on one side lands on receive on the other.
Then the boring one: both ends have to be the same kind of optic. Short-range 850 nm and long-range 1310 nm will not talk to each other, a 100 Mbit module will not talk to a gigabit one, and multimode optics want multimode glass rather than 9 um singlemode.
One more thing that may or may not apply to your Nortels: some switches run a module check at boot and refuse third-party optics outright. Ours never got a confirmed fix beacuse the tests were waiting on a site visit, but the loopback cord was the step that would have split plant from hardware in an afternoon.
One more reason not to replace a module on the strength of an indicator: on this platform family the reported state goes wrong in the other direction too.
I have a stack of three ERS 5520 with SFP ports 1/48 and 2/48 in an MLT. Port 2/48 comes up with OperStatus down and an orange square in the management interface, while the module sits there with a solid green light and traffic crosses the link perfectly normally.
Green light plus frames on the wire means the optics and the glass are doing what they should, and the lie is in the state the stack keeps for that port - on this software train it is the members that are not the base unit that get it wrong. Forcing a fresh base election, or pulling and re-seating the stack cabling, is cheap enough to try, and a later image may well list it as fxied. What buys you nothing is pulling a working SFP because a square in the UI is orange.