Hi,
See this:
http://www.cisco.com/en/US/docs/switches/lan/catalyst4500/15.1.2/XE_340/configuration/guide/vss.html#wp1063735
Dual-Active Detection Using Enhanced PAgP
Port aggregation protocol (PAgP) is a Cisco-proprietary protocol for managing EtherChannels. If a VSS MEC terminates to a Cisco switch, you can run PAgP protocol on the MEC. If PAgP is running on the MECs between the VSS and an upstream or downstream switch, the VSS can use PAgP to detect a dual-active scenario. The MEC must have at least one port on each switch of the VSS.
In virtual switch mode, PAgP messages include a new type length value (TLV) which contains the ID of the VSS active switch. Only switches in virtual switch mode send the new TLV.
For dual-active detection to operate successfully, one or more of the connected switches must be able to process the new TLV. Catalyst 4500, Catalyst 4500-X, and Catalyst 49xx series switches have this capability. For a list of other Cisco products that support enhanced PAgP, refer to Release Notes for Cisco IOS Release at this URL:
http://www.cisco.com/en/US/products/ps6350/tsd_products_support_series_home.html
When the VSS standby switch detects VSL failure, it initiates SSO and becomes VSS active. Subsequent PAgP messages to the connected switch from the newly VSS active switch contain the new VSS active ID. The connected switch sends PAgP messages with the new VSS active ID to both VSS switches.
If the formerly VSS active switch is still operational, it detects the dual-active scenario because the VSS active ID in the PAgP messages changes. This switch initiates recovery actions as described in the "Recovery Actions" section.
Recovery Actions
An VSS active switch that detects a dual-active condition shuts down (by err-disabling) all of its non-VSL interfaces to remove itself from the network, and waits in recovery mode until the VSL links have recovered. You might need to intervene directly to fix the VSL failure. When the shut down switch detects that VSL is operational again, the switch reloads and returns to service as the VSS standby switch.
Loopback interfaces are also shut down in recovery mode. The loopback interfaces are operationally down and not err-disabled.
