On request of the network team our datacenter team provides fiber optic crossconnects between switches. Once these ISLs are in place we can execute a change at a later time to have the interfaces configured and enabled for use in production. Despite the agreement to carefully patch RX/TX strands carefully, the cable needs an RX/TX swap nonetheless and nobody present on location at time of change execution.
One way to have the DC team check RX/TX connectivity is to enable the switch interface when connecting the fiber, but it will immediately take part in spanning tree and other network protocol negotiations, which is strictly prohibited at that time.
Is there a way to enable (Nexus) switch interfaces to allow to verify the optical interfaces signals without interference in network connectivity? I'm thinking of routed interfaces without IP, but maybe there's some other way to allow to check for link connectivity only, and not make it part of network configuration yet.
Its been a while since i have been engaged in optical networking.
An intelligent cross connect system might have the feature you are looking for.
Thanks for your suggestion. The solution I'm looking for should be it a bit more pragmatic. Our DC-team creates Xconnects at one moment (office hours) and network team enables production link after hours.
So isn't there a simple and straightforward solution to make sure RX/TX are correctly aligned when taking into production, something like a 'test interface eth 1/1 fiber-signal' without impact on operation? :-)
you could always look at udld.
Second paragraph down. It is not without a downside.
If a udld condition is found, the port goe into error disable state.
UDLD is a powerful feature, but it doesn't do what I'd like to achieve:
If I can only verify RX/TX during office hours - so link up is guaranteed - I don't need anyone on DC during change execution, as I'm performing change remotely.
Hope this explains the situation...
Not sure of you're environment, but why not just create L3 interfaces on either side and assign temporary IPs that are not included in your routing? (ie 188.8.131.52/30 & 184.108.40.206/30).
This way you'll be able to verify connectivity, there should be no routing protocol peering or neighbor establishment, the network is only known between the two devices and L2 trunking would not come into play. You could also ping across the link to verify link integrity.