Administrator Guide

Stack Master Election
The stack elects a master and standby unit at bootup time based on two criteria.
• Unit priority — User-configurable. The range is from 1 to 14. A higher value (14) means a higher priority. The default is 0. By
removing the stack-unit priority using the no stack-unit priority command, you can set the priority back to the default value
of zero. The unit with the highest priority is elected the master management unit; the unit with the second highest priority is elected
the standby unit.
• MAC address (in case of priority tie) — The unit with the higher MAC value becomes the master unit. The stack takes the MAC
address of the master unit and retains it unless it is reloaded.
To view which switch is the stack master, enter the show system command.
A change in the stack master occurs when:
• You power down the stack master or bring the master switch offline.
• A failover of the master switch occurs.
• You disconnect the master switch from the stack.
When a stack reloads and all the units come up at the same time; for example, when all units boot up from flash, all units participate in the
election and the master and standby are chosen based on the priority or MAC address.
When the units do not boot up at the same time, such as when some units are powered down just after reloading and powered up later to
join the stack, they do not participate in the election process even though the units that boot up late may have a higher priority
configured. This happens because the master and standby have already been elected, hence the unit that boots up late joins only as a
member.
When an up and running standalone unit or stack is merged with another stack, based on election, the losing stack reloads and the master
unit of the winning stack becomes the master of the merged stack. For more details, see sectionsAdd Units to an Existing Stack and
Remove a Unit from a Stack. It is possible to reset individual units to force them to give up the management role or reload the whole stack
from the CLI to ensure a fully synchronized bootup.
Example of Viewing Stack Members
Virtual IP
You can manage the stack using a single IP, known as a virtual IP, that is retained in the stack even after a failover.
The virtual IP address is used to log in to the current master unit of the stack. Both IPv4 and IPv6 addresses are supported as virtual IPs.
Use the following command to configure a virtual IP: Dell(conf)#virtual-ip {ip-address | ipv6–address | dhcp}
Failover Roles
If the stack master fails (for example, is powered off), it is removed from the stack topology.
The standby unit detects the loss of peering communication and takes ownership of the stack management, switching from the standby
role to the master role. The distributed forwarding tables are retained during the failover, as is the stack MAC address. The lack of a
standby unit triggers an election within the remaining units for a standby role.
After the former master switch recovers, despite having a higher priority or MAC address, it does not recover its master role but instead
takes the next available role.
To view failover details, use the show redundancy command.
MAC Addressing on Stacks
The stack has three MAC addresses: the chassis MAC, interface MAC, and null interface MAC.
All interfaces in the stack use the interface MAC address of the management unit, and the chassis MAC for the stack is the master’s
chassis MAC. The stack continues to use the master’s chassis MAC address even after a failover. The MAC address is not refreshed until
the stack is reloaded and a different unit becomes the stack master.
NOTE:
If the removed management unit is brought up as a standalone unit or as part of a different stack, there is a
possibility of MAC address collisions.
772 Stacking