cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
857
Views
12
Helpful
10
Replies

spanning tree issue...help needed

mo shea
Level 1
Level 1

Hi...

I think this is a spanning tree issue. We have a basic setup of One 6509 core switch and 33 2950 switches connecting to it via SINGLE plinks, no redundant paths. Some of the 2950s are stacked together using fiber patch cords and GBICS. We also have one laser

wireless (lightpointe FSO) connected to the 6509 via UTP and the other FSO connects to a single 2950 Fast Ethernet port (UTP). Although the FSO has fiber interface, we use a media converter to connect it to Ethernet ports (as trunk ports) in the switches.

The whole setup was working fine. We switched off the 2950 switch connected to the FSO for 5 months atleast after that for building maintenance. Once we turned on the switch this week and started pinging to our core switch we noticed that in 5 minutes 3 other switches spread around the campus were down. After checking the errors we found that the Gig uplink from those

switches was put in err disable mode and the log shows loopback error. This only happened to switches that were stacked, sometimes the first switch (connected directly to 6509)

and sometimes the second switch (connected to first 2950).

We fixed the problem by switching the GBICS to the othe free slot. We tried to ping again to the FSO switch but another 3 switches (stacked) were down. I do not understand why this is happening, especially because this FSO never gave us any problems like that

before. But this time I also noticed that whenever I telnet to the 2950 connected to the FSO the session is not smooth, more like choppy, which did not happen before. I am suspecting the some configuration was not saved hence lost when this switch was turned off for 5 months.

I wold be grateful for any help and/or some more info on how to track the source of this

loop.

Thanks

1 Accepted Solution

Accepted Solutions

Does the 6509 have a Sup720? If so, what rev of code?

I have witnessed this behavior before on Sup720 Etherchannels. Early verions of IOS for Sup720 were very buggy with Etherchannels, especially across modules. If so, I would run nothing less than 12.2(18)SXD5.

Please rate all helpful posts.

Regards,

Brad

View solution in original post

10 Replies 10

pciaccio
Level 4
Level 4

This has symptoms like a VTP issue. If your VTP domain is not set correctly any switch that has been defaulted or any new switch set for server will take down the network. Also the choppy issue could also be that your default gateways are not set correctly on the switches. I think you need to investigate your VTP and gateway settings. You may find the answer there...Good Luck...Please rate...

Thanks for your reply.

I thought that the use of VTP domain and password should prevet a defaulted switch or vtp server from taking over.

I didnt check the FSO switch because the site is closed, but as far as I can remember the configuration was saved. I will check and confirm this week.

Thanks

Hello,

make sure that the 6509 is the root for all your VLANs, you might want to do this manually with the global command:

spanning-tree vlan x priority 0

where 'x' is the VLAN number; if you have multiple VLANs, configure this command for each VLAN.

Regards,

Nethelper

dgahm
Level 8
Level 8

The 2950 is looping back traffic, but since you have a non-looping topology with the single FSO link to the switch, it is not likely to be a spanning tree issue. What is connected to the 2950? Can you disconnect all user ports on the 2950, and see if it still occurs. What VLAN is the loopback traffic happening on? That may offer a clue.

I have seen loopback Err-Disable occur when dual NIC servers with bridging accidentally enabled are connected via Portfast enabled switchports.

Are you doing any manual VLAN pruning on the trunk port? I have seen an issue with a 3548 where disallowing an VLAN on an access switch caused it loop back the VLAN.

The classic explanation for Loopback Err-Disables is a misconfigured or malfuntioning etherchannel between 2 switches. But it doesn't seem to fit your description.

Thanks for all the responses.

I have had some ideas but couldnt test yet, just waiting when time permits and network is not busy.

To answer dgahm, no manual pruning is being done, only automatic on the 6509. I have disconnected all ports except for the one connecting to the FSO but same result.

While checking my core switch I noticed that the gig ethernet port speed on the 6509 is set to 1000 manual, while the 2950 port connecting to FSO is set to 100. I think this caused the bridge ID for the 2950 to change, but this is just a thought.

I use MST, some vlans are in instance 1, rest are in 2, and made the 6509 as the root bridge (not 0, used priority high which ave the 6509 a 24xxx ID) I did not set any prioriy for MST instance 0, which has vlans that are not defined or used, but I will do that.

May I know how can I check which vlan caused the loopback traffic?

One more thing I noticed in the 2950 connected to the FSO is that the Fe port is both a trunk and an access port, so I deleted the access vlan statement from that port, although it was working perfectly before.

I will test and give my feedback.

Thanks.

Hi...

I tested the setup and hell broke loose. Same problem, GIG uplinks of edge switches going to err disable mode again.

I checked the affected switches and found that some of them were not configured properly for MST, i.e. the vlans were not mapped to instances, unlike other switches. Although this might contribute to the problem, some of the switches that were down had a perfectly configured MST, same region name, rev number, mapped vlans.

Any ideas are greatly appreciated.

Thanks

Even with no Spanning Tree a switch never sends frames back out the same port they were received on. Spanning tree only comes into play when there are multiple links that could be used.

I would suspect the FSO link, or maybe the media converter. Can you plug the 2950 directly into the core 6509 to test that?

Does the 6509 have a Sup720? If so, what rev of code?

I have witnessed this behavior before on Sup720 Etherchannels. Early verions of IOS for Sup720 were very buggy with Etherchannels, especially across modules. If so, I would run nothing less than 12.2(18)SXD5.

Please rate all helpful posts.

Regards,

Brad

Thanks for the response...

I have sup720 as brad mentioned, running 12.2(18) SXD3. I am seriously thinking of upgrading it to 12.2(18)sxf2.

We do not run any ether chnannels at all. All are single links. I am starting to think that some root selection is going on, since no loops can form as there are no redundant links in our topology.

I will have think more throughly before testing the

link again. I will post my results.

Thanks

Spanning tree should only be setup on the Spanning tree server. All VLANs are cerated on the core and propagated out to the edges swithces via membership to teh doamain.

Did you look at spanning tree on all swithces to verify that they are all memebers of the same domain and that the root bridge is your core 6509?

There is a macro (command) you can run on the 6500 to set it to root bridge but I can't remember it off the top of my head.

GL

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco