Switches, Hubs, and Modems
cancel
Showing results for 
Search instead for 
Did you mean: 

Many RSTP root changes

A Dawson
Occasional Visitor

Many RSTP root changes

on several ( but not all my switches ) I see some many RSTP spanning tree root changes

I 01/21/09 10:15:00 stp: RSTP Root changed from 32768:000cdb-48f500 to
32768:000cdb-4d7e00
I 01/21/09 10:15:01 stp: RSTP Root changed from 32768:000cdb-4d7e00 to
32768:000cdb-48f500



Here is a diagram that shows the network topology


HP 9308-1 <--> HP 9308-2
|
|
|
|
|
HP 2650-1 <--> HP 2650-2


There was a second link from 9308-2 to 2650-2, but we removed it to simplify whilst we are investigating some spanning tree problems.


Spanning tree is enabled on the 9308's and on the 2650's

Why do I see all these log entries on only some of the 2600's.

The configurations are similar across our company, on all the switches, but not all show the RSTP changes.
3 REPLIES
Jason Scott
Regular Advisor

Re: Many RSTP root changes

A few things we've learnt recently about spanning tree and HP kit:

Despite a switch claiming to be running RSTP it might be running STP until a reboot. This is seen by ports entering listerning, learning states rather than transitioning to forwarding immediately. This causes problems because by default a port in basic STP will cause a topology change. In RSTP a port is an edge port by default and won't cause a topology change.

Try (switch firmware supporting) the following:

show spanning-tree debug-counters ports [ethernet] instance 0

Look for Topology Change Rx with a last updated time which matches the last spanning tree reconvergence. Use this command to narrow down where the change originated from.
Jason Scott
Regular Advisor

Re: Many RSTP root changes

Sorry, also we've noticed that some of our switches are not seeing topology changes despite running RSTP, having been rebooted and no bpdu-filters being present. We believe there's some bugs in certain firmware which cause this behaviour.
Pieter 't Hart
Honored Contributor

Re: Many RSTP root changes

simply a bad connection can trigger a topology change.
a link with a lot of data traffic also.

Instead of letting the switches decide wich is the root using all standard rules, designate one switches as the spanning-tree root and another as secondary root using the spannning-tree priority.
This will lower the number of topology changes drastically.