Catalyst Swiches in a RING- CONFIGURATION

[bsisco]

As I understood from a previous conversation, earlier IOS versions supported a daisy chain (stacked) backbone connection, while later versions were to support a hub and spoke (using a 3508 for example) and ring patterns (letting SPAN's portfast network loop detection disable one arm). However upon trying this configuration, I only succeeded in failing POST on the switch with the "last leg" connection. Can anyone calrify whether or not a ring pattern is supported by IOS (tm) C3500XL Software (C3500XL-C3H2S-M), Version 12.0(5)WC3b, running on (12)C3524XL-EN switches?

Thanks!

[michael-faust]

I don't see what failing POST has to do with your configuration. What caused the failure (error message)? Will the switch boot if left unconnected? Did it boot properly at one time? Give us some more detail so we can get a better handle on the problem.

Also you talk about 'portfast'. In a ring configuration you are introducing a loop in the spanning tree environment. You do not want to enable 'portfast' on any trunk ports in this case.

[bsisco]

Did it boot properly at one time?

- Yes it was previously in production and therefore booted without issue

What caused the failure (error message)?

-The switch would not boot in the ring config and unfortunately I do not have an error as my laptop was acting up. However the POST stopped after the LED for port 2 went out. (port 3 on were still lit).

Will the switch boot if left unconnected?

- Removing the connection for the ring enabled the switch to finish POST and continue to boot and operate normally.

I understand portfast and what it does, all trunk ports have portfast disabled therefore letting SPAN indentify the loop and disabling the arm creating it.

I will attempt today to ascertain the exact error on POST when in the loop configuration. You're right in that I did not expect the switch to fail at this point either. What in POST would fail with a logical path being the only change?

FYI I also tried using a replacement GBIC to test the possiblility of one being bad I saw the same results.