The newer versions of NX-OS support 1G or 10G on the N5K, but the N5K is not auto-speed sensing. A mini case study follows.
One of my customers just put some some GLC-SX-MM optics and a fiber patch between a 3750 and a N5K. He happened to look at status on Network Assistant on the 3750, and this error appeared:
Description: Gi1/1/1: This port has been disabled because Non Compliant Gigabit Interface Converter (GBIC) connector detected.
Recommendation: Replace connector with cisco compliant Gigabit Interface Converter (GBIC) connector. Refer switch technical documentation to determine cisco compliant connector. Enable the port again.
This seemed odd, since the GLC-SX-MM is supported on both devices. I tried shut/no shut on both ends of the port, but that did not help. I then wondered – does he have a bad SFP? I poked around a bit.
The CLI show log info on the 3750 was also unhappy:
*Apr 16 01:48:41.190: %PHY-4-SFP_NOT_SUPPORTED: The SFP in Gi1/1/1 is not supported *Apr 16 01:48:41.190: %PM-4-ERR_DISABLE: gbic-invalid error detected on Gi1/1/1, putting Gi1/1/1 in err-disable state
Strange, the GLC-SX-MM is supported on the 3750. The interface capabilities looked fine:
3750_sw1#sh int gig 1/1/1 capa GigabitEthernet1/1/1 Model: WS-C3750X-48P Type: 1000BaseSX SFP Speed: 1000 Duplex: full Trunk encap. type: 802.1Q,ISL Trunk mode: on,off,desirable,nonegotiate Channel: yes Broadcast suppression: percentage(0-100) Flowcontrol: rx-(off,on,desired),tx-(none) Fast Start: yes QoS scheduling: rx-(not configurable on per port basis), tx-(4q3t) (3t: Two configurable values and one fixed.) CoS rewrite: yes ToS rewrite: yes UDLD: yes Inline power: no SPAN: source/destination PortSecure: yes Dot1x: yes 3750_sw1#
The status on the 3750 was not connected:
3750_sw1#sh int gig 1/1/1 status Port Name Status Vlan Duplex Speed Type Gi1/1/1 notconnect 1 auto auto 1000BaseSX SFP 3750_sw1#
I then went to look at the N5K. It claimed it has an invalid SFP of an unkown type:
N5K1# sh int e1/5 status -------------------------------------------------------------------------------- Port Name Status Vlan Duplex Speed Type -------------------------------------------------------------------------------- Eth1/5 -- sfpInvali 1 full 10G SFP-1000BAS N5K1# N5K1# sh int e1/5 capa Ethernet1/5 Model: N5K-C5548UP-SUP Type (SFP capable): 10Gbase-(unknown) Speed: 1000,10000 Duplex: full Trunk encap. type: 802.1Q Channel: yes Broadcast suppression: no Flowcontrol: rx-(off/on),tx-(off/on) Rate mode: none QOS scheduling: rx-(6q1t),tx-(1p6q0t) CoS rewrite: no ToS rewrite: no SPAN: yes UDLD: yes Link Debounce: yes Link Debounce Time: yes MDIX: no Pvlan Trunk capable: yes TDR capable: no FabricPath capable: yes Port mode: Switched FEX Fabric: yes N5K1#
The N5K also claimed there was no transceiver:
N5K1#sh log | inc 1/5 ... 2012 Apr 6 18:10:26 N5K1 %ETHPORT-5-IF_DOWN_NONE: Interface Ethernet1/5 is down (None) ... 2012 Apr 6 19:06:10 N5K1 %ETHPORT-5-IF_HARDWARE: Interface Ethernet1/5, hardware type changed to No-Transceiver
I scratched my head a bit, and then came up with the underlying issue – when installing 1GE optics in the N5K, you need to manually set the speed. If you look back, you will see the previous ‘sh int e1/5 status‘ on the N5K shows a speed of 10G. So after the appropriate “speed 1000” command to the e1/5 interface on the N5K, the link came up on both ends.
N5K1# sh int e1/5 stat -------------------------------------------------------------------------------- Port Name Status Vlan Duplex Speed Type -------------------------------------------------------------------------------- Eth1/5 -- connected 1 full 1000 SFP-1000BAS N5K1#
Summary: The newer versions of NX-OS support 1G or 10G on the N5K, but the N5K is not auto-speed sensing. If you run into something similar, even error messages on the remote device, check the port speed on the N5K.
— cwr
Hey Carole, we ran into the same exact issue on our 2232 SFP+ 1G / 10G FEX. If you add a GLC-T Copper GBIC in the slot you get the same "sfpInvali" message until you issue the "speed 1000" command. We selected the 2232 to help us transition from a 1G environment into a 10G Data Center but sadly many of our HP iLO ports and IBM HMC Console ports are 100mb and are not compatible. I added a 3750 for the management links but are soon to add a 2248 to replace it. I’m wondering how many other people are running into these issues. You would think if you spent 500k+ on mainframe servers they would have 1G NICs…