cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1012
Views
5
Helpful
4
Replies

Spanning Tree "Problem"

Willard Dennis
Level 1
Level 1

Hi all,

I have Rapid STP running on all my switches. The (redundant) distribution layer switches are Cisco, and the access layer switches are Dell. When I unplug the link to the root bridge from an access layer switch, it's taking 31 sec's to recalculate the tree before packets start forwarding again. This seems far too long to me... I though the packet loss should only be a few sec's due to RSTP usage. How can I debug why this is happening?

Thanks,

Will

Sent from Cisco Technical Support iPhone App

1 Accepted Solution

Accepted Solutions

Peter Paluch
Cisco Employee
Cisco Employee

Hello Will,

How exactly have tested the speed of convergence? Were you using some end host pinging a different station, and estimating the length of the connectivity dropout?

This seems as if the ports towards end hosts were not configured as RSTP Edge Ports. A topology change in an RSTP network will cause the non-edge ports in the Designated Forwarding role/state to go into Discarding state (a result of a received Proposal). If ports to end hosts are not properly configured as edge ports, they will have to go through Discarding/Learning/Forwarding process which takes 30 seconds.

Please verify on both the Cisco and the Dell switch that all ports towards end stations are indeed configured as edge ports. On Cisco, the setting is done using the spanning-tree portfast interface level command. Unfortunately, I do not know the command for the Dell switch.

If the edge ports are configured correctly then we should inspect the RSTP operation in your network in more detail.

Best regards,

Peter

View solution in original post

4 Replies 4

Peter Paluch
Cisco Employee
Cisco Employee

Hello Will,

How exactly have tested the speed of convergence? Were you using some end host pinging a different station, and estimating the length of the connectivity dropout?

This seems as if the ports towards end hosts were not configured as RSTP Edge Ports. A topology change in an RSTP network will cause the non-edge ports in the Designated Forwarding role/state to go into Discarding state (a result of a received Proposal). If ports to end hosts are not properly configured as edge ports, they will have to go through Discarding/Learning/Forwarding process which takes 30 seconds.

Please verify on both the Cisco and the Dell switch that all ports towards end stations are indeed configured as edge ports. On Cisco, the setting is done using the spanning-tree portfast interface level command. Unfortunately, I do not know the command for the Dell switch.

If the edge ports are configured correctly then we should inspect the RSTP operation in your network in more detail.

Best regards,

Peter

Thanks Peter, that was it... I'm new to the RSTP spanning tree stuff, and forgot to specify the PC port as "Edge". Now it doesn't drop any pings at all (I'm testing by using 1-sec pings from host A to host B through the switches.) It's also spanning-tree portfast on a Dell switch, by the way.

All the inter-switch links are showing as "P2P" as appropriate.

Thanks again for the quick help!

- Will

Will,

Thank you the rating and for letting us know that it worked for you, and also thanks for letting us know about the Dell command

Best regards,

Peter

Jon Marshall
Hall of Fame
Hall of Fame

In addtion to Peter's suggestion can you also check that all your switch interconnect links are defined as P2P and not shared.

Jon

Review Cisco Networking products for a $25 gift card