Re: Spanning tree troubleshoot - possible loop

david-flores · ‎01-19-2006

Hi,

I have several 3548's connected via FEC to my 6509. All these ports are trunking and all setup as channels and configured properly on the 6509 and each 3548. Just yesterday 1 user complained about having issues staying connected to a server. I started to do a continious ping to his system and noticed many time outs. On the switch 3548 that his is connected to, I removed the port group command on the 3548 and the 6509 and this fixed the issue. Could there be a possible loop somewhere on one of the 3548 ports? Need some troubleshooting guidance. Thank you

ToddWarren · ‎01-19-2006

David,

Can you paste show version of 6509 please? I can think of some EC troubleshooting commands depending on this info. Or check out this link to see if it helps.

http://www.cisco.com/en/US/customer/tech/tk389/tk213/technologies_tech_note09186a0080094714.shtml

Todd

david-flores · ‎01-20-2006

WS-C6509 Software, Version NmpSW: 6.3(10)

NMP S/W compiled on Nov 4 2002, 16:46:14

System Bootstrap Version: 5.3(1)

Hardware Version: 2.0 Model: WS-C6509 Serial #:

twojciac · ‎01-19-2006

Etherchannel can definitely be the source of loops if it's misconfigured. You can get 2 ports channeling on 1 side, and 1 on the other, creating a virtual unidirectional link. For this reason, if you must use Etherchannel, use the desirable mode on both sides. Consider utilizing 802.1w instead of Etherchannel if you are looking for redundancy instead of bandwidth, the failover and restore times are roughly 2 seconds.

david-flores · ‎01-20-2006

I don't think it is misconfigured, I have several other 3548's connected to my 6509 configured exactly the same, and I am not having any issues with these switches. The config is exactly the same on all 3548's, as well as the other ports on the 6509. I do have ports on both sides set as trunk ports set to dot1q. I will into 802.1w. I am using these links more as bandwidth than redundancy. Do you have any links for configuring 802.1w? Thanks.

twojciac · ‎01-20-2006

I totally missed the fact that you are running with 3548s. Unfortunately the 3500XLs don't support PAgP, which is what desirable mode requires. You have to use the "on" mode which is unconditional port channeling, which leads to a large number of misconfigured loops.

Make sure your ports and interfaces are configured identically. Even a difference in QoS configs can bring one port out of the channel on one end.

I'd also suggest enabling UDLD to ensure there are no unidirectional links.

http://www.cisco.com/en/US/products/hw/switches/ps637/products_configuration_guide_chapter09186a008007e838.html#xtocid12