C3550 connected to C3560 over 250km - problem with STP

[Konstantin Dunaev]

Hi,

may be someone had experience with a following problem:

there are 2 locations which are connected via SDH ring, they are 200-300 km away from each other.

We're trying to connect C3550 on one site with C3560 on the another site. There is no loop, simply a single trunk connection.

The link comes UP but after some seconds the port on one site (always C3550) goes in block mode with the log message "Loop detected on the interface ...."

after disabling Spanning tree on that interface it went UP and satys active without any problem.

What I can see that on the C3550's interface the "keepalive" is active (I suppose it's default), I didn't try to disable it yet but I think it cause the problem but I can't explain why.

The question: is there any distance (in km or time-responce) limitation for STP?

Thank you!

[Pavel Bykov]

Not distance per se, but delay plays a role.

Distance does have a direct impact on delay, but I have ~250km links as well running with normal PVST without any problems.

Remember, you are running STP to have a loop free L2 network, so if you don't have any loops, and they are not a threat on this connection, you can turn it off without drawbacks.

[Konstantin Dunaev]

Hello,

thank you for the answer.

with delay it's clear, but what I mean is we're _currently_ don't have any physical loop in the design, but we'll defeanetly add some redudency links which could build the loop then. That is why simply disabling the STP is not the option, at least not the best one I suppose.

"no keepalive" should help, may be the migration to RapidSTP will help.

But it's interesting to know the source of that problem in order to solve it in most efficient way.

"loop detected" means that a switch sees its own "keepalive" packet on the same interface, how can it be in such a configuration?