STP topology change

[Hp 5300 Went Down]

Hi, we have a large network (80 switches) and we have a very frequent topology change of the stp.
When we issue the command "show spanning" we notice that the topology change every 40/60 seconds.
There is some command to troobleshoot this problem?

we have an 8212 switch at the core and some 4200, 2524 and 2600 at the edge.

Thank you to all

[idnyc]

Have you tried manually setting your 8200 core switch to a higher root bridge priority?

[Hp 5300 Went Down]

Sure, the core switch (8212) is the root of stp, with priority 0

[Jack Howes]

The best advice I can give you on this type of problem is to look into the device logs to see if the changes are particular to a specific Vlan. Also with so many switches you have to make sure you don't hit the seven switch limit/per spanning tree instance. Of course this is supposition on my part since we don't know what your topology looks like.

[Matt Hobbs]

The most important thing is that you make sure they're all running the current firmware versions. 9 times out of 10 that will stabilise it.

[RicN]

Jack:

>Also with so many switches you have to make
>sure you don't hit the seven switch
>limit/per spanning tree instance.

What is this seven switch limit? How does it work and what importance does it have?

[Jack Howes]

Normal spanning tree has a limit of seven hops from the end/bottom switch to the root of the spanning tree per Vlan. It then depends on if your switches can run per vlan spanning tree or one spanning tree for all of your vlans. If the BPDU's are not received from the root in a timely manner, a device assumes there is a problem and swithes to the other uplink and STP change. Again without details I'm assuming most of your switches are dual homed.

[RicN]

Thank you for your answer, Jack!

>Normal spanning tree has a limit of seven
>hops from the end/bottom switch to the root
>of the spanning tree per Vlan.

With normal spanning tree, do you mean the original STP standard or does this "problem" exist with RSTP and MSTP also?

>If the BPDU's are not received from the
>root in a timely manner, a device assumes
>there is a problem and swithes to the other
>uplink and STP change.

If I am not incorrect the TTL for BPDU:s from the root is 20 seconds, if that is the time-limit for the bpdu:s to cross the network from root to "leaf" switches - will there be a problem with modern high performance switches?

(With that I mean, the seven switches limit, is that a "hardcoded" limit or a soft limit defined to be sure that bpdu has enough time to travel over the network?)

[Matt Hobbs]

That's an old limit from the original spanning-tree specification. I believe it's now up to 40 hops with MSTP.

[Jack Howes]

Yes, the seven hop limit is for STP, not for RSTP or MSTP. I tried to find an actual limit for RSTP on Cisco Web site and other resources and couldn't find one.

Have you been able to narrow down what is causing your frequent STP changes? Are all your switches running RSTP or MSTP? Do you think you have a spanning tree loop of some kind or just a flakey link that is causing the topology changes?

[RicN]

Thank you Matt and Jack! One more reason for not running the legacy-spanning tree protocol.

(I did not post the original question however.)