I have 2 new 3750's connected via GBICs that work fine. When I used a cross-over cable as the manual suggests to connect the 3750 to the normal network switch which is a 2900, the light starts, then goes black. The web based management piece shows on the port "Status: Port in error-disabled".
The recommended solution is that a GBIC is bad, but since I can bring up both systems and communicated back and forth, I believe they are good.
The other suggestion is enabling the spanning tree or something. I can telnet and configure this if that's what's necessary, but I'd like the step by steps if possible.
Other than that, I find it hard to believe that these Ciscos wouldn't talk to each other natively, is there something else I'm missing?
I also downloaded the Cisco Network Assistant which runs very slow...selected all ports and chose enable. Would this have the same affect as enabling the port in Telnet for the spanning-tree?
Help, I'm really at a loss.
Thanks for any help...Jake
Please do a show log on your switch and it will give the reason why the port is int error disable state. Please paste that output here so that we can carry forward with the reason.
Also let us know that the cross cable is fine and tested on other switch. Please paste the config from both the ports that you are inter-connceting. Also did you try chaging the speed and duplex on the ports. Please try to hardcore the speed and duplex on the ports also.
Thanks for your quick response, I'm waiting for this system to log in now. Taking way too long The show log displays in numerous ports comments such as:
PM-4-ERR_DISABLE: bpduguard error detected on Gi1/0/22, putting Gi1/0/22 with BPDU Guard enabled. Disabling port.
...changed state to down
Looks like the above problem is what's going on, now that I know to do a show log. I tested with another known working Cross-over cable same result.
I found some info yesterday on performing this BPDUGUARD and enabling it, but I wasn't sure when I was at the telnet screen what the preceding commands would be, such as advanced, or enable, then the command, then save or write or something. This part of the cisco stuff is just too new for me.
I did not try changing any of the default speeds or duplex mode, but again, now it's really pointing to this BPDUGUARD problem I have.
What do you think?
This is the normal and the expected behaviour with the BPDUGUARD fetaure. This feature is working as expected. We enable BPDUGUARD on the ports to avoid loops on the network. When we have BPDUGURAD enable on the ports and it sees a BPDU on the port from another switch it puts the port in error disable state to avoid the loops in the n/w.
What happened in your case is the BPDUGUARD was enable on your switchport and the moment you connected the 2900 on that port it sees the BPDU from the neighbor switch and put its port in error disable state.
What you can do to connect two switches right now is that you disable the BPDUGUARD on the port and it will work fine.
bpduguard only kicks in when a port has 'portfast' turned on. Portfast should be off, i.e., in default spanning-tree mode, between two switches.
bpdu guard is doing exactly what it should be doing in this case. Leave it on. It's portfast you want to turn off.
Also, for future reference, if you want to trunk between these switches, keep in mind that the XL series switches don't support DTP, so you have manually trunk.
IMHO, you should not run BPDU-guard or portfast on any inter-switch link, regardless of whether you are trunking. Do you have these features enabled on the switchport? Beware that they might be enabled by global default, in which case you should soecifically disable them on the inter-switch links.
There are a variety of problems including a faulty NIC (as opposed to the GBIC itself being faulty) that could cause this problem. On the switch end, it is probably a port configuration issue.
Here is a useful document on recovering from errDisable port state on CatOS platforms: http://www.cisco.com/en/US/tech/tk389/tk214/technologies_tech_note09186a0080093dcb.shtml.
Otherwise forget CNA, it may have some really cool features but its way too slow to be of much use... :-(
It seems like a spaning-tree issue. I suggest you review stp configurations of each switch, especially bpdu guarding...
I think it is a Duplex issue. Change the interface speed and duplex from auto to 100 and full duplex on both switches and see what happens. I think it should work.
Also, if trunking between switches old 2900xl's do not understand ISL. It must be a DOT1Q trunk. And, do you have errordetection enabled? You can do a show errordetection to see why the port is being error disabled if that is still the case.