[c-nsp] Migrating from STP "pathcost short" to "pathcost long"

[Peter Rathlev]

Hello everybody,

We've been migrating from 16-bit pathcost values ("pathcost short",
Cisco default) to 32-bit values ("pathcost long") a couple of places and
I can't seem to figure out if there are supposed to problems in a
transition state.

We have seen some strange symptoms, like no L2 connectivity between
devices in the same VLAN. Chronology points at pathcost inconsistency
being the cause, though I can't see why.

When migrating, there will of course be short periods where the STP
topology is strange, but after reconvergence I don't see why long and
short can't co-exist in a network.

Would anybody know if we're supposed to see these problems? Is it a
"no-no" to mix short ang long pathcosts in a spanning tree?

Thank you in advance.

-- 
Peter