Background:
I have two office locations, each with a Cisco 3560G switch. It is desireable to have both sites within the same LAN segment (well technically there are a few VLANs, but that is beside the point for now). The locations two switches were initially connected to one another via fiber, using an ethernet to fiber media converter on each end, which formed a transparent L2 bridge between the switches. Everything has been working well for quite some time as if the switches were plugged into one another with a single piece of UTP cable.
Because of the necessity of this link to remain available, an alternate link between the two sites was obtained in the form of a T1 circuit -- it is a point to point private line, nothing from the telcos in the way. What I have done with this is connect it to a T1 interface card in a Cisco 28xx series router on each end, and set them both up to bridge the Serial interface to an Ethernet interface. This has essentially given me another transparent L2 bridge, which works just like the fiber link (with less bandwidth of course).
The plan now is to connect the two 3560's together, and have the link via T1 remain in a blocked state, using spanning tree to manage the "failover" as necessary if the fiber link were to go down. However, when I connect both links, I get a loop in the network and a resulting broadcast storm until one is disconnected (or at least shutting down an interface on one end).
I was under the impression that as long as the BPDUs go through (which they do) the switches should be able to properly negotiate via STP which port(s) to block & which to forward regardless of the up/down state of the interface.
Relevant configuration information:
Building A
----------
spanning-tree mode pvst
spanning-tree extend system-id
spanning-tree vlan 1-3,37,100 priority 24576
!
interface GigabitEthernet0/18
description Link to B via T1
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-3,100
switchport mode trunk
!
interface GigabitEthernet0/24
description Link to B via fiber
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-3,37,100
switchport mode trunk
spanning-tree port-priority 64
Building B
----------
spanning-tree mode pvst
spanning-tree extend system-id
spanning-tree vlan 1-3,37,100 priority 28672
!
interface GigabitEthernet0/6
description Link to A via fiber
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-3,37,100
switchport mode trunk
spanning-tree port-priority 64
!
interface GigabitEthernet0/7
description Link to A via T1
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 1-3,100
switchport mode trunk
Official Questions:
Is there any reason that the redundant links between switches have to all be direct physical connections (as in a single piece of cable)?
Is there some additional STP configuration that has to be used because of the fact that the switch interface does not go "down" when a fiber cut occurs?
Is the problem that I'm using PVST (which allegedly only works with ISL) instead of PVST+ (which does work with 802.1Q, but doesn't seem to be an option on the 3560)?
Is there something else that I'm missing here?
P.S. Yes, I intentionally left VLAN 37 out of the T1 link because that would be carrying SAN volume replication traffic, which I do not need bogging down the T1 during a failover scenario. The replication will be queued up while the primary link is restored, this is fine.
Start Free Trial
