MISTP configured but not so rapid: Convergence time still 50 Seconds.

[albert_coll]

Hello, I configured two 7606 running 12.2(18)SXD3 each. They are connected via an aggregated etherchannel trunk configured to be the root port or designated port depending on the different MST instances defined.

Due to some special requests, i inserted a hub (cisco 1835) with a link to an access port of every 7606 in vlan 10. I setup the designated port of the nonroot 7606 switch with a higher cost to become in blocking state. MISTP converged as expected.

The problem is that if i disconnect the link from hub to the root bridge (which was in forwarding state), then, the designated port wich was in blocking state takes 50 seconds to be in forwarding, that is, the usual transition time of 802.1d.

2) I issued the command “spanning-tree link-type point-to-point” in every two ports, because the ports connected to the hub are in HALF duplex, and the switch assumes that half-duplex links are shared links and full-duplex links are point-to-point links but rapid connectivity is established only on point-to-point links.

Unfortunately, the convergence time is still 50 seconds.

Does somebody know anything else to do ?

Regards in advace

Albert.

SWITCH A:

I include port settings and MIST settings.

!

spanning-tree mode mst

no spanning-tree optimize bpdu transmission

spanning-tree extend system-id

!

spanning-tree mst configuration

name xxx

revision 1

instance 1 vlan 2-499

instance 2 vlan 500-999

!

spanning-tree mst 0-1 priority 24576

spanning-tree mst 2 priority 28672

!

interface GigabitEthernet4/1

no ip address

logging event link-status

logging event spanning-tree status

switchport

switchport access vlan 4

switchport mode access

spanning-tree link-type point-to-point

spanning-tree mst 0-1 cost 20000

spanning-tree mst 2 cost 30000

SWITCH B:

spanning-tree mode mst

no spanning-tree optimize bpdu transmission

spanning-tree extend system-id

!

spanning-tree mst configuration

name xxx

revision 1

instance 1 vlan 2-499

instance 2 vlan 500-999

!

spanning-tree mst 0-1 priority 28672

spanning-tree mst 2 priority 24576

!

interface GigabitEthernet4/1

no ip address

logging event link-status

logging event spanning-tree status

switchport

switchport access vlan 4

switchport mode access

spanning-tree link-type point-to-point

spanning-tree mst 0-1 cost 30000

spanning-tree mst 2 cost 20000

!

[Francois Tallet]

Hi Albert,

RSTP/MST will not help you a lot here, however I think it should be a little bit faster.

When the link between the root and the hub fails, the alternate port stops receiving bpdus. As it does not detect the link going down, it will age out the information in 3xhello_time = 6 seconds by default.

Then this port will move from alternate to designated discarding. As it is not receiving any kind of agreement, it will take an additional 2xforward_delay = 30 seconds to go forwarding. I rather expect 36 seconds instead of 50 seconds in your case. Please make sure that the port is not in STP compatibility mode (that would be surprising.. do "show span mst" and check there is not "Bound(STP)" in the type column).

The only way to be faster would be to tune the forward delay at that stage, I don't recommend changing the hello time (not worth it). I don't think you'll be much faster unless you can replace your hub by a bridge running at least RSTP.

Also, there is something I need to mention about your setup. Be careful that MST will consider this access port as a backup link between your two core switches for instance 0 and 1. That's probably not an issue because it seems you don't rely on STP for redundancy between your core switches, but you should be aware of it.

Regards,

Francois

[albert_coll]

Francois,

Thank you for your answer. I'll take in account your comments.