A Failed Chassis in an VSB Stacking System
If a chassis fails, there may be traffic disruption lasting several seconds as the system processes link down events and discovers the new topology.
Be aware that it is possible for a stack to be partitioned into separate operational segments. Should this happen, each segment operates as though it is a separate whole stack using the same IP address. The closed ring configuration will prevent separate partitions when a single point of failure occurs by assuring a daisy chain exists for all remaining chassis in the stack. A closed ring configuration is configured using interconnect links between the first and last chassis in the VSB stack.
Topology changes occur when a replacement chassis first joins the stack. Some traffic disruption may accompany these topology changes.
To replace a failed chassis in an VSB stacking system:
- Remove the failed chassis taking note of the switch/routing, interconnect and uplink cabling that currently exists.
- Before enabling VSB on the replacement chassis, assure that the replacement chassis has been loaded with the same firmware image loaded in the current system.
- The configured bonded chassis-ID is associated with the physical chassis in memory. Clear the memory of the failed unit from the stack before configuring bonding for the replacement unit using the clear bonding chassis failed-chassis-ID command.
- Configure the replacement chassis‘ VSB chassis ID and the ID of the VSB system to agree with the configuration of the failed chassis.
- Optionally, on a two chassis VSB stacking configuration, enable bonding on 10GbE monitor ports.
- Cable the replacement chassis with the cable configuration noted in step 1.
- Restore the port level configuration by appending the configuration with the configuration from a previously stored configuration file when the chassis was operational within the stack, using the configure filename append command.
- Enable bonding on the replacement chassis‘ 40GbE interconnect ports.
- Globally enable the VSB system. The system is reset.
Note
If the bonded MAC address on each chassis is different (as provisioned by the network administrator), a host master election occurs resulting in a reset of all chassis except the elected host master.