Aruba 2540 uplink ports 51-52 stay Down on an inter-building run while a 2530 links the same modules
I look after a two-building site and the access switches were replaced with Aruba 2540s a while back. The fibre between the buildings has not been touched, but since the swap the uplink refuses to come up.
- Aruba 2540 on our side, ports 51 and 52 used for the uplinks
- Aruba J4858D 1000SX in the 2540, TP-Link TL-SM311LM v2.0 on the far end of that pair
- on a second pair, Aruba J9050D 10G against a TP-Link TXM431-SR
- one inter-building run, patched at both ends, nothing else changed
The switch reads the modules without complaining:
show tech transceivers
51 1000SX J4858D
52 1000SX J4858D
port 51: Down
port 52: DownWhat I have already ruled out:
- same modules, same fibre, a 2530 in place of the 2540 - link comes up
- TP-Link to TP-Link over the same run - link comes up
- 2540 to 2540 with J9050D on that fibre - still Down
- both modules on a short patch cord on the bench - link comes up
So the fibre carries a link for every combination except one that ends in a 2540, and show log -r gives me nothing beyond the port going down. What is the 2540 doing differently here?
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 3
If everything links except the box that is a 2540, stop looking at the optics and pin the rate. Those cages negotiate the speed themselves, and against third-party gear with rate-selectable modules that negotiation is exactly where these links stall. It also fits your bench result: over a short cord the far end settles fast enough, over the long run it does not.
Set it explicitly on the uplinks:
Do one port first and watch
show log -rwhile it comes up, then the second. If the far end has a fixed 1000 full option, set that too rather than leaving one side negotiating.Two caveats. This is what fixed the same symptom on a site I worked on, not something I can point to in a manual, so prove it on one port of your own kit before you roll it out to every 2540. And formally none of the TP-Link modules are supported in an Aruba switch - if this ever becomes a support case the first request will be J4858D at both ends. If the fixed speed changes nothing, moving that particular run back onto a 2530 is a perfectly defensible answer until you have a supported pair of modules on both sides.
Post the full
show tech transceiversfrom the working 2530 and from the failing 2540 side by side. What I am after is whether the 2540 reads those modules exactly the way the 2530 does, or whether it is reacting to something the older box quietly ignores.The other thing to rule out on the 2540 is spanning tree protection on those ports. BPDU protection, root guard or loop protect will pull a port down within a second of it coming up and from the far end it looks like a dead link. Run
show log -randdebug eventwhile you seat the module - that tells you whether the port ever came up at all or never saw light.Both boxes read the module the same way.
show tech transceiversgives me 1000SX, J4858D on ports 51-52 on the 2530 and on the 2540, line for line - nothing in that output tells the switch that liinks apart from the one that does not.No protection features on those ports either, they are plain uplinks with a VLAN and nothing else.
show log -rshows the port going down and nothing before it, anddebug eventgives me no transceiver events at all while the module is seated.The TP-Link sidde is out of the picture anyway: two 2540s with J9050D at both ends of the same fibre also stay Down.