Users Guide

Failure scenarios
On a link failover, when a VLT port channel fails, the trac destined for that VLT port channel is redirected to the VLTi to avoid
ooding.
When a VLT switch determines that a VLT port channel has failed (and that no other local port channels are available), the peer
with the failed port channel noties the remote peer that it no longer has an active port channel for a link. The remote peer then
enables data forwarding across the interconnect trunk for packets that would otherwise have been forwarded over the failed port
channel. This mechanism ensures reachability and provides loop management. If the VLT interconnect fails, the VLT software on the
primary switch checks the status of the remote peer using the backup link. If the remote peer is up, the secondary switch disables
all VLT ports on its device to prevent loops.
If all ports in the VLT interconnect fail, or if the messaging infrastructure fails to communicate across the interconnect trunk, the
VLT management system uses the backup link interface to determine whether the failure is a link-level failure or whether the remote
peer has failed entirely. If the remote peer is still alive (heartbeat messages are still being received), the VLT secondary switch
disables its VLT port channels. If keepalive messages from the peer are not being received, the peer continues to forward trac,
assuming that it is the last device available in the network. In either case, after recovery of the peer link or reestablishment of
message forwarding across the interconnect trunk, the two VLT peers resynchronize any MAC addresses learned while
communication was interrupted and the VLT system continues normal data forwarding.
If the primary chassis fails, the secondary chassis takes on the operational role of the primary.
The SNMP MIB reports VLT statistics.
RSTP and VLT
VLT provides loop-free redundant topologies and does not require RSTP.
RSTP can cause temporary port state blocking and may cause topology changes after link or node failures. Spanning tree topology changes
are distributed to the entire layer 2 network, which can cause a network-wide ush of learned MAC and ARP addresses, requiring these
addresses to be re-learned. However, enabling RSTP can detect potential loops caused by non-system issues such as cabling errors or
incorrect congurations. To minimize possible topology changes after link or node failure, RSTP is useful for potential loop detection.
Congure RSTP using the following specications.
The following recommendations help you avoid these issues and the associated trac loss caused by using RSTP when you enable VLT on
both VLT peers:
Congure any ports at the edge of the spanning tree’s operating domain as edge ports, which are directly connected to end stations or
server racks. Disable RSTP on ports connected directly to Layer 3-only routers not running STP or congure them as edge ports.
Ensure that the primary VLT node is the root bridge and the secondary VLT peer node has the second-best bridge ID in the network. If
the primary VLT peer node fails, the secondary VLT peer node becomes the root bridge, avoiding problems with spanning tree port
state changes that occur when a VLT node fails or recovers.
Even with this conguration, if the node has non-VLT ports using RSTP that you did not congure as edge ports and are connected to
other Layer 2 switches, spanning tree topology changes are still detected after VLT node recovery. To avoid this scenario, ensure that
you congure any non-VLT ports as edge ports or disable RSTP.
VLT Bandwidth Monitoring
When bandwidth usage of the VLTi (ICL) exceeds 80%, a syslog error message (shown in the following message) and an SNMP trap are
generated.
%STKUNIT0-M:CP %VLTMGR-6-VLT-LAG-ICL: Overall Bandwidth utilization of VLT-ICL-LAG (port-
channel 25)
crosses threshold. Bandwidth usage (80 )
When the bandwidth usage drops below the 80% threshold, the system generates another syslog message (shown in the following
message) and an SNMP trap.
%STKUNIT0-M:CP %VLTMGR-6-VLT-LAG-ICL: Overall Bandwidth utilization of VLT-ICL-LAG (port-
channel 25)
reaches below threshold. Bandwidth usage (74 )VLT show remote port channel status
906
Virtual Link Trunking (VLT)