LevelOne GTL-2691 - Manual (Page 87)

C HAPTER 2 | Initial Switch Configuration
Stack Operations
– 87 –
S ELECTING THE
B ACKUP U NIT
Once the Master unit finishes booting up, it continues to synchronize
configuration information to all of the Slave units in the stack. If the Master
unit fails or is powered off, a new master unit will be selected based on the
election rules described in the preceding section. The backup unit elected
to serve as the new stack Master will take control of the stack without any
loss of configuration settings. To ensure a logical fail over to the next unit
down in the stack, place the Slave unit with the lowest MAC address
directly beneath the Master unit in the stack.
R ECOVERING FROM
S TACK F AILURE OR
T OPOLOGY C HANGE
When a link or unit in the stack fails, a trap message is sent and a failure
event is logged. The stack will be rebooted after any system failure or
topology change. It takes two to three minutes to for the stack to reboot. If
the Master unit fails, the backup unit will take over operations as the new
Master unit, reboot the stack, and then select another backup unit after the
stack finishes rebooting. If a unit is removed from the stack (due to a
power down or failure) or a new unit added to the stack, the original unit
IDs are not affected after rebooting, and a new unit is assigned the lowest
available unit ID.
B ROKEN L INK FOR L INE AND W RAP -A ROUND T OPOLOGIES
All units in the stack must be connected via stacking cable. You can
connect the units in a simple cascade configuration from the top to the
bottom unit. Using this kind of line topology, if any link or unit in the stack
fails, the stack will be broken in two. The Stack Link LED on the unit that is
no longer receiving traffic from the next unit up or down in the stack will
begin flashing to indicate that the stack link is broken.
When the stack fails, a Master unit is selected from the two stack
segments, either the unit with the Master button depressed, or the unit
with the lowest MAC address if the Master button is not depressed on any
unit. The stack reboots and resumes operations. However, note that the IP
address will be the same for any common VLANs (with active port
connections) that appear in both of the new stack segments. To resolve the
conflicting IP addresses, you should manually replace the failed link or unit
as soon as possible. If you are using a wrap-around stack topology, a single
point of failure in the stack will not cause the stack to fail. It would take
two or more points of failure to break the stack apart.
N OTE :
If a stack breaks apart, the IP address will be the same for any
common VLANs (with active port connections) that appear in both stack
segments.
R ESILIENT IP I NTERFACE FOR M ANAGEMENT A CCESS
The stack functions as one integral system for management and
configuration purposes. You can therefore manage the stack through any
IP interface configured on the stack. The Master unit does not even have to
include an active port member in the VLAN interface used for management
access. However, if the unit to which you normally connect for
management access fails, and there are no active port members on the