I'm trying to resolve an STP problem which has taken our whole network down the last two days, and I want to make sure I have the ideal MSTP configuration on our Catalyst 6509 for interoperating with all our switches, especially Meraki but also some legacy HP.
Initially I had this configuration on our core switch when I transitioned it from PVST to MSTP:
spanning-tree mode mst
spanning-tree extend system-id
spanning-tree mst 0-1 priority 8192
spanning-tree vlan 1-4094 priority 8192
We're a school district, and I set this up during the summer when very few staff were in the buildings. Reading up on MSTP there were suggestions about creating regions and instances, so I added this:
spanning-tree mst configuration
name region1
instance 1 vlan 3-4, 6-8, 12-16, 100, 110, 120, 200, 300, 999-1000
I didn't see effects, positive or negative, to adding this. Are they necessary or helpful in any way, or should I just remove the region and instance to the core switch?
For further mitigation today I added "spanning-tree guard root" to all ports on the Catalyst 6509 and added this to the config:
spanning-tree portfast edge bpduguard default
I'm hoping this will help prevent spanning tree from going totally haywire again. Today and yesterday at around 10am STP topology changes became so rapid on the network that most of the switches only saw themselves as root and the Meraki switches got in such a state that they needed a full reboot even with almost everything unplugged from them so they could even respond normally to pings on their admin interfaces. On advice of Meraki support I administratively disabled the redundant switch trunk links at multiple sites with the idea being it would simplify the tree (which is 6 switch layers deep) and hopefully keep convergence time under control going forward. We'll LAG/bond those links in the future, and we're also planning to move some of the layer 3 switch routing to inside buildings and change their uplinks from trunks to access to further simplify the spanning tree topology.
