STP on CE500

[phlitservices]

Hi Guys,

I'm down to my last straw with my current LAN design/layout. Any assistance will be appreciated. I have 3 Catalyst Express 500 switches and 1 3COM unmanaged switch. All of my servers are connected to the 3com. Attached is my current troublesome design which is causing loops all around my network (i can see when i use ethereal) . Please let me know where/if I should enable STP. I tried enabling it on the CISCOs and it caused the GB uplink port CiscoSW00 to be blocked.

Any assistance is appreciated.

Thanks

Phil

[jackyoung]

Why you said it is looping ? According to the diagram, the all switches are cascaded one-by-one. Could you please provide another diagram which show the loop ?

Moreover, you have to define the root STP to define where is the blocking point.

Check here for STP info. :

http://www.cisco.com/en/US/tech/tk389/tk621/technologies_configuration_example09186a008009467c.shtml

Hope this helps.

[phlitservices]

Hi,

Thanks for your reply. First, when I run ethereal I have one interesting event that repeated roughly 15 times in about 20 seconds.

" Source Dest. Prot. Info"

Cisco_71:43:93 Cisco_71:43:93 LOOP Reply

Also I can see a lot of DHCP Request packets being dropped in between. Clients sometimes don't get addersses from the DHCP server. The network is also very slow. This problem only started this today after the switches were installed.

Thanks

[jackyoung]

Could you please confirm the switches are physically looped or not ? And where ?

Please also provide the config. of CE500 ?

Is there any VLAN in the design ? Where is DHCP server located ?

I need more info. for analyze. Thx.

[phlitservices]

Hi Again,

The answers to your questions are as follows:

1) The switches are not physically lopped.

2) There are no additional VLANS, only the default VLAN1

The DHCP server is connected to the 3COM switch. (All of the servers are connected to the 3COM switch. All clients are connected to the CISCO switches.)

Attached is the config for the 3 CE500s, You can also take a look at the previously posted Diag. to clarify things.

Thanks again.

Phil

[jackyoung]

Thanks for the config. I have below questions & comments :

1) What device is connecting to the Ciscosw0 FE port 24 ? 3COM switch ? Because I found it is trunk mode;

2) It is fine to set portfast at user connection, recommend to configure normal SPT mode at the port which is connecing to other switches;

3) If there is no VLAN, better to configure the cascaded port (e.g. GE) to access port instead of trunk mode;

4) Can you provide the "sh cdp nei" at the Cisco switches ?

5) The config. looks fine, if there is no physical looping, and no VLAN routing, it should not be looped.

6) Confirm the speed/duplex is matching w/ all equipment and hosts, e.g. from 3com to Cisco, from server to 3com, from user to cisco.

Please try to connect one user to the 3com and determine the DHCP issue and performance.

Hope this helps.

[Francois Tallet]

The switch may not be physically looped, but may be they are attached to some hosts that are doing some bridging and that can introduce a loop.

I definitely recommend you keep STP enabled on the Cisco devices, even if you don't need it in theory. If, when enabling STP, you have a blocked port, it means that STP detected a loop. It would be much easier to start the troubleshooting from there because it will help locate where the loop is.

If the network is exactly as you describe it, then there is no loop. A loop is not detected by the LOOP messages that you saw with ethereal, but the symptoms you mention (slow network etc...) are indeed the ones of a possible bridging loop. The fact that STP blocks a port when enabled is definitely to be investigated.

Regards,

Francois