Testing CWDM SFPs and CWDM-MUX-8A muxes back-to-back after links went err-disabled
2
In the poster's data center several Cisco CWDM SFP/GBIC links stopped working: some ports went err-disabled (one with link-flap as the cause) and others simply showed notconnect. Troubleshooting included shut/no shut, reseating the modules, moving the CWDM SFPs to other ports, replacing the patch cords and checking the interface media type, but every port ended up notconnect. Suspecting the CWDM-MUX-8A multiplexers, the poster wanted to build a lab with two spare muxes in a point-to-point setup to rule out the optics and muxes, and asked about optical power levels, minimum fiber length, how to loop a Catalyst interface for testing, and whether an OTDR could be used on the mux monitor port in production.
- Hardware: Cisco Catalyst switch (model unspecified)
- Transceivers: Cisco CWDM SFP/GBIC modules (wavelengths unspecified), CWDM-MUX-8A multiplexer
- Form factor: SFP, GBIC
Symptoms / messages:
port err-disabled (link-flap)
notconnect
several CWDM links failed at once
Comments 3
For two CWDM-MUX8-A muxes back-to-back, insert about 10 dB of attenuation: +5 dBm max TX minus -7 dBm max RX means 12 dB of loss is needed, the two muxes contribute only 2 x 2.2 dB, leaving 7.6 dB to add. CWDM receivers are wideband, so any CWDM wavelength can be received by any CWDM transceiver.
Loop each CWDM module's TX to RX with a patch lead and an attenuator (TX around +3 dBm versus a -7 dBm maximum RX); use 'no errdisable detect cause loopback' so the Catalyst does not err-disable the looped port. Two same-wavelength CWDM optics can also be linked back-to-back with attenuators, since the receivers accept 1450-1620 nm.
Transmitting from one SFP into another SFP on the same switch will likely show link without harming the receiver; refer to Cisco's SFP troubleshooting guide. Using an OTDR on the mux monitor port of a live network is unwise because its broadband ~1550 nm pulse may cross into active channels.