Extreme Networks 200 Series - Manual (Page 244)

242 Summit 200 Series Switch Installation and User Guide
Configuring Stacked Switches
Recovering a Stack
Whenever the stack is active, the stack master monitors the stack members for link state changes, such
as a link changing from up to down. However, the master monitors the stacking links at all times for
changes in stack topology. Examples of a change in stack topology are a switch being added, deleted, or
a link being down that results in the lost of connectivity to a member switch.
When the master detects there is a link state change on a stacking port, it restarts the Stack Discovery
protocol. Loops in stacking ports are detected and automatically cut during Stack Discovery, however,
this configuration is not supported.
The master monitors the stack topology by sending out a heartbeat command approximately once every
two seconds to each switch in the stack. The master can then detect any changes to the stack and update
any data structures as needed. Any changes are flagged in the log. Depending on the type of error
detected, either an entire slot can be deemed unusable or stacking might be disabled. The heartbeat also
detects any loops in the stacking ports.
If the stack is broken because stacking is disabled; a stack link goes down; or a switch in the stack goes
down, the following occurs on the master and the stack members.
• Master —If stacking is still enabled, the master continues to function as a stack of n (where n is the
number of switches the master can access). The heartbeat process continues to try and reconnect to
the remaining switches. You can still enter configuration commands including configuration
commands for the slots that are no longer accessible. When the stack comes back up, the master
resynchronizes with the stack members and sends down any new commands entered on the master
to each switch that is out of date.
• Members —Members wait for the stack master come up. If the master does not boot, the software
uses the timeout value in the config stack slave timeout [none | <n>] to determine whether
to reboot the member. If the default value of none is configured, the switch operates as a non-stacked
switch. To communicate with the member, either connect through the console port or if the StkMgmt
VLAN was configured, Telnet to the unique IP address for the particular switch. The members
continue to attempt to reconnect through the stacking ports. If a timeout value is specified, the
member reboots using either the primary or secondary saved configuration after the indicated
amount of time.
In cases where the stack does not automatically recover, you might want to disable or unconfigure
stacking without reconfiguring the stack. If you want to unconfigure stacking on the master without
rebooting all the member switches to the original configuration, use the
disable stacking command
as described in “Changing a Stack Configuration”.
To reboot all of the member switches and clear their configuration in the master switch, enter the
following command:
unconfigure switch all
To reboot the member switches and to revert the members to their previously selected configuration,
enter the following command:
unconfigure stacking
If the unconfigure stacking command is issued on the stack master when the stack is enabled, the
command passes to all switches in the stack. The master saves the configuration without stacking
information and reboots. All switches revert to non-stacked switches. If the
enable stacking