D-Link DXS-3600 Series Reference Manual page 86

DXS-3600 Series Layer 3 Managed 10Gigabit Ethernet Switch Web UI Reference Guide
Backup Master - The Backup Master is the backup to the Primary Master, and will take over the
functions of the Primary Master if the Primary Master fails or is removed from the Stack. It also monitors
the status of neighboring switches in the stack, will perform commands assigned to it by the Primary
Master and will monitor the running status of the Primary Master. The Backup Master can be set by the
user by assigning this Switch the second highest priority before physically assembling the stack, or it can
be determined automatically by the stack through an election process which determines the second
lowest MAC address and then will assign that switch as the Backup Master, if all priorities are the same.
The Backup master are physically displayed by the seven segment LED to the far right on the front panel
of the switch where this LED will flash between its given Box ID and 'h'.
Slave - Slave switches constitute the rest of the switch stack and although not Primary or Backup Masters,
they can be placed into these roles when these other two roles fail or are removed from the stack. Slave
switches perform operations requested by the master, monitor the status of neighbor switches in the
stack and the stack topology and adhere to the Backup Master's commands once it becomes a Primary
Master. Slave switches will do a self-check to determine if it is to become the Backup Master if the
Backup Master is promoted to the Primary Master, or if the Backup Master fails or is removed from the
switch stack. If both Primary and Backup masters fail, or are removed from the Switch stack, it will
determine if it is to become the Primary Master. These roles will be determined, first by priority and if the
priority is the same, the lowest MAC address.
Once switches have been assembled in the topology desired by the user and powered on, the stack will
undergo three processes until it reaches a functioning state.
•
Initialization State - This is the first state of the stack, where the runtime codes are set and
initialized and the system conducts a peripheral diagnosis to determine each individual switch is
functioning properly.
•
Master Election State - Once the codes are loaded and initialized, the stack will undergo the
Master Election State where it will discover the type of topology used, elect a Primary Master and
then a Backup Master.
•
Synchronization State - Once the Primary Master and the Backup Master have been
established, the Primary Master will assign Stacking Unit IDs to switches in the stack,
synchronize configurations for all switches and then transmit commands to the rest of the
switches based on the users configurations of the Primary Master.
Once these steps have been completed, the switch stack will enter a normal operating mode.
Stack Switch Swapping
The stacking feature of the Switch supports "hot swapping" of switches in and out of the running stack.
Users may remove or add switches to the stack without powering down or largely affecting the transfer of
data between switches in the stack, with a few minor provisions.
When switches are "hot inserted" into the running stack, the new switch may take on the Primary Master,
Backup Master or Slave role, depending on configurations set on the newly added switch, such as
configured priority or MAC address. Yet, if adding two stacks together that have both previously
undergone the election process, and therefore both have a Primary Master and a Backup master, a new
Primary Master will be elected from one of the already existing Primary Masters, based on priority or MAC
address. This Primary Master will take over all of the Primary Master's roles for all new switches that were
hot inserted. This process is done using discovery packets that circulate through the switch stack every
1.5 seconds until the discovery process has been completed.
The "hot remove" action means removing a device from the stack while the stack is still running. The hot
removal is detected by the stack when it fails to receive heartbeat packets during its specified interval
from a device, or when one of the stacking ports links is down. Once the device has been removed, the
remaining switches will update their stacking topology database to reflect the change. Any one of the
three roles, Primary Master, Backup Master or Slave, may be removed from the stack, yet different
processes occur for each specific device removal.
If a Slave device has been removed, the Primary Master will inform other switches of the hot remove of
this device through the use of unit leave messages. Switches in the stack will clear the configurations of
the unit removed, and dynamically learned databases, such as ARP, will be cleared as well.
75