Brocade FastIron Ethernet Switch Stacking Configuration Guide - Manual (Page 96)

Traditional stack topologies
Brocade traditional stack technology supports both linear and ring stack topologies. Because the
unicast switching follows the shortest path in a ring topology, this topology offers the strongest
redundancy. When the ring is broken, the stack recalculates the forwarding path the resumes the flow
of traffic within a few seconds. In a ring topology, all stack members must have two stacking ports,
however, In a linear topology, both end units use only one stacking port, leaving the other port
available as a data port. To see an illustrated example of each topology, refer to “Brocade traditional
stacking topologies.”
Port down and aging
If a unit is powered down, or the stacking link is removed, the system immediately detects the port
down and knows that its neighbor is gone. That unit is immediately removed from the active controller.
If a unit is gone or no longer stack-enabled, but its stacking link is still on, it will take 20 seconds to age
the neighbor out. The following message will be logged and displayed.
Warning! my mac=0000.004f.4243, age out up-stream
Traditional stack device roles and elections
There are three distinct roles played by units that are part of a traditional stack:
• Active controller
• Standby controller
• Stack member
Active controller
The active controller contains the saved and running configuration files for each stack member. The
configuration files include the system-level settings for the stack, and the interface-level settings for
each stack member, as well as MIB counters and port status. The standby controller also has a
synchronized copy of the active controller startup config file for use in the event the active controller
fails.
When a stack is formed, the console function for each stack member is automatically redirected to the
active controller console. The active controller console port handles all stack management functions,
as well as ping, Telnet sessions, and TFTP image downloads for every stack member. If you connect
to the console port on a stack member that is not the active controller, you are automatically directed
through the console of the active controller.
The active controller synchronizes its start-up configuration with the standby controller and the rest of
the stack members. You can recover the previous flash configuration of the standby controller and the
stack members by issuing the stack unconfigure command. For an example of this command and
the output generated, refer to
Unconfiguring a traditional stack on page 61.
The active controller may reset the rest of the stack members, if necessary. However, if the active
controller itself must be reset because of a role or ID change, you must issue the reset command.
If the active controller fails, the standby controller waits 30 seconds, and then takes over as active
controller, resetting itself and all other stack members. If the old active controller becomes operational,
it may or may not resume its role as active, depending on the configured priorities. If hitless stacking
failover is enabled, the standby unit can take over immediately without reloading any unit.
Traditional stack topologies
94 FastIron Ethernet Switch Stacking Configuration Guide
53-1003090-02