Cisco Catalyst 2960 Series Switches

Cisco Catalyst 2960-S FlexStack - Description, Usage, and Best Practices White Paper

  • Viewing Options

  • PDF (1.6 MB)
  • Feedback


Cisco ® FlexStack stacking for Cisco Catalyst ® 2960-S Series Switches provides a true stacking solution with all switches in a FlexStack stack acting as a single switching unit. FlexStack provides a unified data plane, and single configuration for a group of stacked Cisco Catalyst 2960-S switches.
FlexStack lowers the total cost of ownership with built-in 1:N redundancy, high availability, and preprovisioning, not available in standalone switches. High-availability features such as EtherChannel and FlexLinks will work across stack members, increasing uptime and network connectivity.
As part of the Borderless Network architecture, FlexStack allows scalable provisioning and management to deploy any number of switches quickly and easily.

Why Stack Ethernet Switches?

Stacking Ethernet switches reduces the network administrator's total cost of ownership. The cost of maintaining the network is decreased because there are fewer devices to manage, and the network uptime is increased with built-in redundancy.
Figure 1 shows two ways to deploy Ethernet switches at the access layer. In each case four switches are connected to a Cisco Catalyst 6509 Switch. The deployment scenario on the left shows four switches that are colocated in the same wiring closet, but are not stacked. Each standalone switch on the left has two uplinks to the distribution layer switch that are combined in an EtherChannel group. The four switches on the right are stacked together with FlexStack (FlexStack links show in red). The stack has four uplinks to the distribution layer switch. Because FlexStack supports EtherChannel grouping of Ethernet interfaces across the stack members, the four uplinks are grouped together into a single EtherChannel group. All the benefits of EtherChannel grouping are retained when switches are placed into a stack.

Figure 1. FlexStack Comparison

Stacking Ethernet switches provides the network administrator with three major operational benefits:

• Single point of management: All switches in the stack are managed as one.

• Built-in redundancy and high availability: The high-speed FlexStack connections provide redundant communication for each stack member to every other member.

• Scalable to fit network needs: Installation of a new switch to the stack is easy. As the need for additional access ports grows, adding a new switch to an existing stack is easier and faster than adding a new standalone switch to the network.

The first operational benefit is fewer devices to manage. Multiple physical switches in a stack appear as a single logical switch. This eases management overhead because there are fewer devices in the network to manage. A single IP address is used to manage the logical switch. All manageable entities (for example, Ethernet interfaces and VLANs) on all physical switches can be configured and managed from the logical switch. The logical switch will appear as a single entity in the network. In a Layer 2 network, the logical switch will appear as a single spanning-tree entity.
The second operational benefit that stacking provides is built-in redundancy and increased availability. Data path redundancy is built into the stacking architecture as there are two physical paths between any two stack members. Connecting stack members with the stacking cables provides data path redundancy for all stack members. Stacking increases Ethernet switch availability by providing redundancy for both the physical switch and the uplink. Because different physical switches will connect to the upstream network, losing one switch or one uplink interface does not prevent connectivity to the network. Since the logical switch has multiple uplinks, the logical switch still has network connectivity because at least one uplink is still active.
In a logical switch, a single physical switch is acting as the stack master. The stack master manages all physical switches, including itself. If the master fails, another member automatically becomes the stack master following a well-documented election process (covered later). The configuration of the stack is preserved through a single switch failure or a reboot of all stack members. FlexStack provides 1:N redundancy for the stack master, with the ability for any physical switch to back up the acting master. (See Figure 2.)

Figure 2. FlexStack with Redundant Links

Flexibility of FlexStack allows for modular stacking capability. Cisco Catalyst 2960-S switches can be added to a stack at any time by adding the optional FlexStack module. As the demand for Ethernet access ports increases, Cisco Catalyst 2960-S switches can be stacked together either to create a new stack or to add new Cisco Catalyst 2960-S switches to an existing stack. This flexibility provides investment protection for the network administrator, who can add additional stack members as needed.
Adding members to a stack and replacing physical units are much easier when multiple switches are combined into a FlexStack group. Since the stack retains the configuration, there is no need to back up the configuration prior to removing the switch. Conversely, when a new member is inserted into the stack, the configuration of the stack is pushed to the new member. The network administrator need not explicitly recover the switch configuration since the stack took care of it.

Stacking and Clustering

Stacking is not clustering. Clustering is a technology available on Cisco Catalyst fixed Ethernet switches that allows the network administrator to use a single public IP address to manage multiple physical switches. Clustering was created to assist network administrators by having a single point of management as well as in preserving valuable public IP addresses needed to manage the Ethernet switches. Stacking is much more than a single point of management and IP address preservation because it offers redundancy, availability, and ease of management. Cisco Catalyst 2960-S supports clustering as well as stacking.

What Is FlexStack?

FlexStack is the name of the stacking technology used by the Cisco Catalyst 2960-S series of fixed Ethernet switches. FlexStack is specific to Cisco Catalyst 2960-S switches, and only Cisco Catalyst 2960-S Series Switches use FlexStack.
FlexStack is composed of hardware components and the FlexStack protocol. The hardware components are the FlexStack module (Figure 3) and the FlexStack cables. The FlexStack module supports two FlexStack ports. The FlexStack module is inserted into the rear of the Cisco Catalyst 2960-S Series Switch. Two FlexStack cables, inserted into the FlexStack module, are used per Cisco Catalyst 2960-S switch to provide data path redundancy. Using FlexStack cables, the physical members of the stack form a ring, providing the built-in redundant data path for each member of the stack.
The FlexStack protocol allows the FlexStack members to behave as one logical switch. The FlexStack protocol runs on each member. The FlexStack protocol is used by the master switch to manage the other members. The master tracks the presence of each member, the member's availability, the Cisco IOS ® Software image, and the status of each member's FlexStack connections.
Stacking with FlexStack on the Cisco Catalyst 2960-S Series Switches is optional. In order for the switches to stack, a separate FlexStack module must be inserted in the rear of the switch. Without this module, FlexStack cannot function. Figure 3 shows a partially inserted FlexStack module in the rear of the Cisco Catalyst 2960-S Series Switch. The FlexStack module is hot swappable.

Figure 3. FlexStack Module

A single FlexStack connection between two members is a full duplex 10Gbps connecton. Each Cisco Catalyst 2960-S member supports two FlexStack connections. Each FlexStack member can simultaneously send and receive Ethernet traffic over both stack links at line rate, effectively giving 20Gbps of stack bandwidth per member.
Up to four switches can stacked together using FlexStack to form a single logical switch. Logical switch is another term used to identify a stack of switches. There are multiple physical switches acting as a single unit. This single unit behavior is a logical switch.
Cisco Catalyst 2960-S switches running FlexStack use a hop-by-hop method of transferring Ethernet packets across the stack. Packets traverse the stack by going from one member to another over the FlexStack links until the packet reaches its destination. This is the same behavior as multiple standalone Ethernet switches forwarding packets from one switch to another.

FlexStack Cabling

Special cables are used to connect Cisco Catalyst 2960-S Series Switches together in a stack. These special FlexStack cables can only be used to stack Cisco Catalyst 2960-S switches with other Cisco Catalyst 2960-S switches. Figure 4 shows a picture of two FlexStack cables connected to a fully inserted FlexStack module. Notice how the tabs on the cables are on opposite sides of each other. The FlexStack cables are keyed as well. Besides the tabs needing to be away from each other, the keying of the metal portion of the connector prevents the cable from being connected incorrectly.

Figure 4. FlexStack Module with Cables

When each member in the FlexStack group has two operational FlexStack links, then the stack is operating in a fully redundant mode. The dual FlexStack connections from each member to two other members are what provide the redundancy. Should any single FlexStack connection break or cease to operate, then the switches in the stack will use the remaining FlexStack connection that is provided.

Deployment Topologies

Figure 5 shows a three-member stack with full bandwidth and with redundant FlexStack connections. Each member in Figure 5 has 20Gbps of stack bandwidth and is fully redundant.

Figure 5. Fully Redundant Three-Member Stack

When one FlexStack Link is not present, the stack still functions. Figure 6 shows a stack with incomplete FlexStack cabling. In Figure 6, all data traffic passes through the middle member. This stack is operating in a nonredundant mode. This stack provides only half of the possible bandwidth between members and does not have redundant connections. Only the middle member has 20Gbps stack bandwidth. The top and bottom members are operating at 10Gbps stack bandwidth.

Figure 6. Nonredundant Three-Member Stack

Three different FlexStack cable lengths are available, shown in Table 1.

Table 1. Cable Lengths

Product ID




0.5 meters

This is the default cable that ships with the FlexStack module.


1.0 meters



3.0 meters


The different lengths allow for deployment flexibility. Figure 7 shows how to stack four switches using only the 0.5-meter cables. The stack member connections are interweaved using the stack cables. No cable extends more than two stack members. The interweaving of the stack links still provides redundant connections for the stack.

Figure 7. Four Member Stack with 0.5-Meter Cables

Figure 8 shows the conventional cabling deployment. The 3.0-meter cable is used to complete the redundant ring connection by connecting the top member with the bottom member. The other connections connect to directly adjacent members using the 0.5-meter cables. The 3.0-meter cable is not the default cable shipped for FlexStack and must be ordered. See the Cisco Catalyst 2960-S data sheet for ordering information.

Figure 8. Four Member Stack with 3.0-Meter Cable

The 0.5-meter cable can be used to connect two Cisco Catalyst 2960-S switches that are 4 rack units away from each other. In a stack of four, with all four members racked directly on top of each other, 0.5-meter cables can be used to connect all stack members. When stack members are more than 4 rack units apart, then longer stack cables are required.

FlexStack Protocol

FlexStack protocol allows every stack member to be in constant communication with the adjacent member and with the stack master. Each member is aware of the operational state of every stack port in the stack. The FlexStack protocol allows the stack to react to changes in the operational state of a FlexStack connection (for example, a stack changes state to operationally down). Changes in a stack port's operational status are pushed to all members as soon as the state change is detected. FlexStack protocol is used to detect new member additions as well as member removal. All stack operational status for members and interfaces is communicated to each member through the FlexStack protocol.
Because all management of the stack takes place on the stack master, member status and stack port connectivity information comes from the stack master too.

Role of the Stack Master

The stack master controls the configuration and is the central point for management. All Layer 2 protocol traffic (for example, VLAN Trunking Protocol [VTP], Dynamic Trunking Protocol [DTP] , Cisco Discovery Protocol, Link Layer Discovery Protocol [LLDP]) is forwarded to the master regardless of where the protocol packet ingresses. The master will also transmit all Layer2 protocol packets. If the egress interface is on another member, then the protocol packet is passed from the master along the stack to the destination member.
The same behavior is seen for management traffic. Simple Network Management Protocol (SNMP), HTTP, Telnet, and Secure Shell (SSH) Protocol type of management traffic is all forwarded to the master regardless of the ingress interface. The response from the master is sent through the stack to the destination interface.
On configuration changes, the stack master pushes a copy of the configuration to every member. This way all members have a copy of the saved configuration.
Stack upgrades occur on the master. The master pushes new Cisco IOS Software images to all members. Each member stores a copy of the Cisco IOS Software image on its local Flash.

FlexStack LED Operation

The LEDs on the front of the Cisco Catalyst 2960-S can be used to view stack operation. When multiple Cisco Catalyst 2960-S switches are stacked together, only the stack master will have the "MSTR" LED solid green. Other members of the same stack will have this LED dark.
To find the switch member number, press the mode button just below the LEDs until the stack LED goes green. At this point the LEDs over the Ethernet ports are used. The port number matching the member will start to blink. For member 2, port 2 will blink. For member 3, port 3 will blink. Depending upon the number of members in the stack (up to four) the other ports will go solid green. This allows the network administrator to see the number of stack members just from the LEDs over the ports.
When a switch is operating standalone, the MSTR LED will be solid green.

FlexStack Details

Inserting the FlexStack Module

The FlexStack module will insert easily into the slot. Just before it is fully inserted there will be "minor" resistance. This is normal. Push the module completely into the switch chassis. Tighten the screws no more than "finger" tight. The screws are meant to keep the module from slipping out, not to maintain stack link connectivity. Finger tight is sufficient to maintain a fully operational stack connection.

FlexStack Redundant Power

Redundant power for each Cisco Catalyst 2960-S is provided by the Cisco Redundant Power System 2300 (Cisco RPS 2300). Each stack member must have its own connection to the Cisco RPS 2300.

FlexStack Operational Management

Managing FlexStack switches involves adding members to and removing members from an existing stack. This section will cover techniques for safely and efficiently adding and removing members from a stack of Cisco Catalyst 2960-S switches.

FlexStack Operations

To see FlexStack operation and member status, use the Cisco IOS Software CLI command "show switch." This command gives the basic information, including number of members, which member is the master, and state of all members. Here is a sample output of the command:
C2960S-48# show switch
Switch/Stack Mac Address : 0022.bdc4.1d80
H/W Current
Switch# Role Mac Address Priority Version State
*1 Master 0022.bdc4.1d80 14 1 Ready
2 Member 0022.bdc4.2300 1 1 Ready
3 Member ec30.912e.2380 1 1 Ready
4 Member 0026.0ac1.3e00 1 1 Ready
The preceding example shows that member 1 is the master. The master has stack priority of 14. All other members have the default stack priority of 1. The state of all members is ready.
Other possible stack member states are:

• "progressing": booting up.

• "mismatch" :a hardware or software mismatch is preventing the member from fully joining.

• "provisioned": this member has been configured, but it is not part of the stack and has never been part of the stack.

• "removed": if there are stack members that are not currently powered on or were removed from the stack, these members would have the state "removed."

The MAC address of the master is used as the MAC address of the logical stack. This is the base MAC address from the master, and it will be used for all IP communications, as well as Cisco Discovery Protocol data exchange and Spanning Tree Protocol.
To view other operational aspects of the stack, these commands are useful:

Show version: view the Cisco IOS Software version of all stack members. Useful if there is a mismatch.

Show switch neighbors: view which members are connected to which stack ports.

Show switch stack-ports: state of the stack ports on each member.

To determine stack port connectivity and operational status, use the commands " show switch stack-ports" and " show switch neighbors." Refer back to Figures 5 and 6. One shows a three-member stack with a full redundant mode. The other shows a stack in nonredundant mode. Here is what the show commands will show for each stack, respectively.
Fully redundant three-member stack:
The master of the stack will be the one management interface for the entire stack. All IP communications (that is, Telnet, SSH, SNMP, ping, HTTP, and so on) will be handled by the master regardless of the member the communication ingressed the stack. All console activity regardless of media (RJ45 or USB) is directed to the master regardless of the member the console port is connected to.
The Cisco Catalyst 2960-S supports the CISCO-STACKWISE-MIB for managing switch stacks. This is the same MIB that the Cisco Catalyst 3750, 3750-E, and 3750-X Series Switches support.

Making a Specific Switch the Stack Master and Master Election Rules

Stack Election Rules

When multiple physical members are present in the stack, one of the members will be elected as the stack master. There can only be one master and up to three backups 1:3.
The election rules used by FlexStack in choosing a master are applied in this order:
1. Switch that is the current master
2. Switch with higher priority
3. Switch that has configuration file
4. Switch with highest uptime
5. Switch with lowest MAC address block
When new switches join an existing stack, then rules 1 and 2 will be applied.
When two standalone switches are joined together, having no switch priority configured, and both having existing configuration files, then the switch with the longest uptime will be chosen.
To have a specific member elected as master, two methods can be used.
1. Set the priority of the desired member to be higher than all other members. This is rule 2 above.
a. Example of configuring member 1 to have priority 14:
C2960S-48(config)#switch 1 priority 14
b. A reboot of the current master will be required after this to change the active master. Just configuring the priority will not by itself change the active master. Save the configuration before rebooting.
2. When the switches boot, and after they have completed power-on self-test (POST), there is a delay of approximately 2 minutes while the switches wait for all switches that could be in the stack to boot. This 2-minute delay is called the election window. Network administrators can take advantage of this window by first powering up only the desired master switch. Once the election window closes, this switch will become the master of a stack of one. Any subsequent switches that boot after the first switch will become members of the existing stack.
The switch with the lowest MAC address is used to break the tie in those scenarios in which the other rules cannot be applied.
It is desirable for network administrators to make a specific switch in the stack the master. This could be the topmost switch or the bottommost switch. For switch identification purposes knowing how the members are connected is an advantage. If a switch needs to be replaced, then it is very convenient to know where that switch is located in relation to the other stack members.

Note: All the Cisco Catalyst 2960-S LAN Base switches are functionally equivalent. All Cisco Catalyst 2960-S LAN Base switches are equally capable of becoming the stack master. Physical place in the stack does not affect switch behavior.

For brand new switches, with no configuration, all powered at the same time, the switch with the lowest MAC address will be the master. It is not practical to orient and deploy switches in a stack based on MAC address. To create a new stack with the correct physical member as the master, the network administrator must use the "uptime" rule. To reliably make any new switch the master, it must be powered up first. Subsequent switches should not be powered up until the election window is complete and the first switch has become the master. This is about 2 minutes, or when the system LED and the MSTR LED turns solid green.
To change the master after the stack of Cisco Catalyst 2960-S switches has been created and set up, use the global configuration command "switch X priority Y," where X is the member number that is desired to be the master, and Y is the priority value to use. To make member X the master, you must assign it a priority value higher than that of the current master.
By default, all switches have priority 1. To view switch priority, use the "show switch" command:
C2960S-48#show switch
Switch/Stack Mac Address : 0022.bdc4.1d80
H/W Current
Switch# Role Mac Address Priority Version State
*1 Master 0022.bdc4.1d80 1 1 Ready
2 Member 0022.bdc4.2300 1 1 Ready
3 Member ec30.912e.2380 1 1 Ready
4 Member 0026.0ac1.3e00 1 1 Ready
To make sure that member 1 will become the master, the priority should be changed to any value greater than 1. Here is an example showing member 1 assigned priority 14:
C2960S-48#config term
Enter configuration commands, one per line. End with CNTL/Z.
C2960S-48(config)#switch 1 priority 14
If a switch other than the master is given higher priority, as shown in the preceding command, there will not be an immediate switchover of the master. The master will not change until the current master is rebooted forcing a stack master election of remaining members.
Setting the priority is important if in the future two stacks will be merged. Existing stacks are merged when the FlexStack cable from one stack is connected to another stack. It only takes one cable to merge two stacks. When two existing stacks are merged, a master election takes place. When two stacks merge, both stacks already have a stack master, so the first rule is not applied. The best way to determine which stack master will continue to be master of the new combined stack is with the priority value.

Note: A standalone switch is still a stack of one member. Adding a standalone switch that has been powered up to an existing stack is a still a stack merge. If the standalone switch has a higher priority than the master of the other stack, then the standalone switch will become the master of the combined stack. During a stack merge the new master is chosen during the election process. The election process is used during a stack merge and at stack power-on time.

As always during a stack merge, new members are rebooted after the election process has completed.

Configuring a Backup Master

It is a good practice to configure another switch to be the backup should the primary master fail. Configuring a backup master makes a switchover predictable in case of primary master failure. To configure a secondary (or backup) master, assign a priority value less than that of the primary master but greater than that of the other members. In the example just given, the primary master has priority 14, and all other members have priority 1. Any member will be secondary master with a value greater than 1 and less than 14.

Determining the Member Number

For brand new switches the order in which switches are powered up initially is also the order the member numbers will be assigned. Brand new switches do not have a member number and default to member 1. As new switches join an existing stack, the new member is assigned the next available member number. When the second switch is powered on, it is assigned member number 2. When the third switch is powered on, it is assigned member number 3. And so on.
If new members are not powered on in any order, then member numbers can appear to be randomly assigned. Understanding FlexLinks connections between members can potentially be confusing when the member numbers are not in an order that matches physical layout.
It is possible to change member numbers of switches in a stack. It is not required for stack operation. The stack will operate just fine when member numbers do not match the physical order of the stack members. In spite of this, it is desirable to have member numbers match the physical layout to ease managing the stack and its members.
Use the Cisco IOS Software CLI to change member numbers. The switches must be rebooted for the member number change to take effect. The following example will renumber member 4 to member 3:
Switch(config#) switch 4 renumber 3
Caution: Changing member numbers causes any configuration on the interfaces of the old member to be lost. All interface configurations are based on the member number. When the member is changed, the configuration for current members is NOT changed or copied to the new member number. In the preceding example, the configuration for all of the Ethernet interfaces on member 4 will be lost, and the configuration for member 3 is unknown. Take care when changing member numbers and be prepared to reconfigure the interfaces.
Each member retains its own member number through a reboot or a power cycle. When a member boots into a stack, the member announces its member number to the stack. If there is no conflict, the member number is retained. If there is a conflict with two different physical members having the same number, then the first member to join the stack retains its number, and the newest member must change its number. The next available member number is assigned to the switch in such a conflict situation.
Since members retain their member number through a reboot, when a member moves from one stack to another, the potential for member number conflict exists. The ability to modify the member number is critical to resolve member number conflicts.

Adding Members to an Existing Stack

The relative ease with which new members can be dynamically added to an existing stack is a real strength of FlexStack. A brand new switch added to an existing stack will join as a member. The new member will be given the next available member number. When a new member is added to a stack, it will be rebooted to clear any configuration it might have. When the new member completes the reboot it will have the configuration of the stack.
It is recommended that FlexStack operate in a fully redundant mode. In order to add a member to an existing stack, at least one FlexStack link must be disconnected, causing the existing stack to operate in nonredundant mode. This disconnection will cause a minor traffic outage across the stack. Traffic will resume while the stack is in nonredundant mode. The network administrator does not need to connect the FlexStack link in a fully redundant mode for traffic to flow. When the FlexStack links return to a fully redundant mode, there will be another minor traffic outage across the stack. Network administrators should try to limit the number of FlexStack link operational changes to decrease the amount of traffic disruption.
Cisco Catalyst 2960-S switches do not need to be powered off to join a stack. However, when a new member joins an existing stack, the new member will be rebooted anyway to clear its configuration . The new member is automatically rebooted by the stack. A reboot is operationally equivalent to powering a switch off and then on again. Thus, it is a better practice to add new stack members to an existing FlexStack powered off.
The following are steps for adding a new member to an existing stack, and it is desired that the new member not become the master. In a rack deployment, new switches are typically added at the top or the bottom of an existing FlexStack. This example will show how to add switches to the bottom of an existing stack of switches.
1. Place the new switch in its final location. Do not insert the power cord. To minimize downtime, complete all cabling that would be required of any switch placed into production at this time.
2. Insert the FlexStack module in the back of the new member.
3. Disconnect the FlexStack cable from the existing member just above to the new member. This would be the cable that is connecting the bottom member with the top member. You will need this to connect the new member located at the bottom of the stack to the member at the top.
a. When the FlexStack cable is disconnected, this is when the stack enters into a nonredundant mode.
4. Connect the new member with a FlexStack cable to the member just above. The FlexStack port used to connect one member to another is not important. Stack members should be connected to each other any way that makes cable management easiser for the network administrator.
5. Connect another FlexStack cable to the member at the top. Make sure that all cables are connected properly. A light tug on the FlexStack cable will make sure that it is properly seated.
6. Connect the AC power cord to the new member. It will power on. It will take a few minutes for the new member to complete the boot process.
7. The new member will join the existing stack.
a. When the new member completes the boot process, this is when the stack enters into a fully redundant mode once again.
8. All the FlexStack links will be operational if all the connections were made properly.
If all went well, traffic on the exiting members was minimally interrupted. There were at least two interruptions. The first occurred when the FlexStack cable was removed in step 3. The other interruption occurred when the new switch joined the stack after being powered on (step 7). In between the two interruptions, traffic will continue to be switched over the existing members and the operational FlexStack links.

Removing Members from an Existing Stack

FlexStack remembers all members that have been members. When a member is physically removed or just powered down, FlexStack still considers that switch a member, just not currently active. The reason that the stack remembers members that have been removed is because most of the time the member comes back. For example when a member is powered off, then powered back on, the member will rejoin the existing stack, and all the member-specific configuration will be reapplied when it completes the boot process. This is the most common occurrence.
If the removal is permanent, then its recommended that the stack configuration be "cleaned up." The member will be deleted from the stack configuration. Here is an example of a command to remove stack member 4 from the configuration after it has been physically removed from the stack:
C2960S-48(config)#no switch 4 provision
If the network admin chooses not to "clean up" the configuration after a member has been removed, the configuration will remain in the running-config and the startup-config for the stack. This extra configuration will make the configuration file "config.text" larger than it should be.
In a switch replacement situation, replacing a member switch with an identical model is very convenient in FlexStack. FlexStack remembers the member model number, but not the MAC address. During a switch replacement, if the old member is replaced by the same Cisco Catalyst 2960-S model, when the new switch is inserted into the FlexStack group, it will inherit the configuration of the previous member. If the new switch model does not exactly match the replaced switch, the old configuration is not used. If there is room in the stack for a new member, then the new switch is added to the stack. If the stack size is at the maximum, then the configuration for the replaced member is removed.
Replacing a master switch needs one additional item to be clarified. The switch priority (configured with the command "switch X priority Y") is not stored in the running-config or the startup-config. When the replacement switch is placed into the stack, it will not inherit the switch priority. The new switch will have the default priority of 1. To make the new switch the master, the network administrator must configure the new switch to have a higher priority than any other member. The existing master must be rebooted (or the entire stack must then be rebooted) after this configuration to force a master change. Just configuring the priority to the new member is not enough to trigger a master change. If the entire stack was rebooted, the newly added switch with the highest priority will be the stack master. If only the master is rebooted, the existing members will have a master election. The switch with highest priority will become stack master. The old master will reboot and join the stack as a member.

Upgrading the Cisco IOS Software Release on a Stack

FlexStack has the capability to upgrade all members with a single command. A single command eases the operational burden on the network administrator. The command to initiate an upgrade takes place on the master. The master makes sure that each member is properly upgraded.
The Cisco IOS Software CLI command to upgrade a FlexStack stack
"archive download-sw [options] <source-file>."
The default behavior is to upgrade all members.
The options for the ` archive download-sw' command allow for detailed control of the upgrade. Use the options to update only a single member instead of the entire stack or to reload the stack after the upgrade is successful instead of having the reboot be an extra step for the network admin.
The "source file" is a tar file downloaded from called an image file. The tar file can be copied over the network using Trivial File Transfer Protocol (TFTP) or FTP. The most common and recommended way to upgrade the stack of switches is using TFTP. The network admin should copy the tar file with the desired Cisco IOS Software image to a TFTP server in the network. In the `archive download-sw` command, use the IP address of the TFTP server and the name of the image file as the source file. No destination file name is necessary.
Here is an example of using the ` archive download-sw` command to upgrade a stack of Cisco Catalyst 2960-S switches. Notice that the number of members in the stack is not relevant to this command. By default all members are upgraded unless the network admin uses the options to do otherwise. The TFTP IP address is, and the tar file name is c2960s-universalk9-tar.122.53.SE.tar
Switch# archive download-sw tftp://
If a network connection is not possible, the stack can be upgraded by reading the image file from a USB Flash drive. A USB Flash drive must be inserted in the USB A slot on the front of the switch. The tar file containing the Cisco Catalyst 2960-S Cisco IOS Software image must be on the Flash drive. The USB Flash drive is referenced by the member it was inserted into. If the USB Flash driver was inserted into member 2, then it will be referenced as "usbflash2:." Use the same download command used in the preceding example, except this time point to the USB Flash instead of the TFTP server:
Switch# archive download-sw usbflash2:c2960s-universalk9-tar.122.53.SE.tar
The Cisco IOS Software CLI command "archive download-sw ..." is very flexible and yet simple to use. It can retrieve source images from multiple locations (for example, FTP, TFTP, Flash). The command allows for detailed control of the upgrade process. For more information about the "archive download-sw" command, see the Cisco Catalyst 2960-S software configuration guide on

Preprovisioning Members

FlexStack allows for preprovisioning of future members. This allows the network administrator to configure Ethernet interfaces for members that have yet to join. In those instances where the network administrator knows how certain interfaces will be configured, the admin can do so in advance to ease the insertion of the new member into the stack and into the network. This is useful when all interfaces that will connect to end users will have the same configuration for every stack member. When new members are added to the stack, if they have been preprovisioned, then it is not necessary for a network administrator to return to the stack and configure the new interfaces.
As an example, the network administrator knows that interfaces 49 and 50 on a future member will be connecting to another Ethernet switch and will be used as uplinks. Interfaces 1-48 will be connecting to end user desktops. The network administrator can make sure that VLANs will match the upstream switch by configuring the Ethernet interfaces 49 and 50 for the proper VLAN membership. The network administrator preprovisions the member number by matching the member number with a supported switch model. Ideally the admin would choose the next available member number, but can choose any available member number.
This example shows how to add a fourth switch to an existing stack of three. The next available member number is 4 based on the output from "show switch." After preprovisioning the member and matching the model, the admin can configure interfaces Gi4/0/1-Gi4/0/48, Gi4/0/49, and Gi4/0/50. The following are the CLI commands to implement this preprovisioning. After the switch has been preprovisioned, all interfaces on the future member can be configured.

Recommended Deployment Strategies for Building FlexStack

These are the important items to consider when deploying FlexStack:

• Configure the stack so that the uplinks are present on at least two stack members.

• Use cross-stack EtherChannel to load balance the traffic on the uplinks.

• If there are only two physical links, and EtherChannel is not desired, then use the backup interface feature.

• When using 10 Gigabit Ethernet as the uplinks, separate the 10 Gigabit Ethernet uplinks across stack members if possible.

– It is always recommended to have at least two uplinks from the stack to the aggregation device.

– This is standard best practice for any deployment regardless if stacking is present or not. It is recommended for physical link redundancy.

• It is recommended that the stack master not contain uplinks if it is possible to distribute the uplinks across other stack members.

– This recommendation should only be followed in three- or four-member stacks.

– If the master fails and it has the uplinks, then those uplinks are lost as well. If the master fails, and it does not have any uplinks, this will cause the minimal amount of traffic loss.

– Configure one member to be the backup master. Use the Cisco IOS Software CLI configuration command "switch <member number> priority <1-15>" to configure a backup. In case of master failure, having a deterministic backup is useful.

• When using only two uplinks in a stack of four members, put the uplinks on members that are not directly connected with FlexStack links. Directly connected members are those connected together with FlexStack connections. This is only relevant on FlexStack groups that contain four members.

– This recommendation is purely for stack bandwidth efficiency. Placing the FlexStack uplinks on directly connected members is less efficient on stack bandwidth. In a four-member stack, with members connected 1-2, 2-3, 3-4, 4-1, put the uplinks on members 2 and 4 or 1 and 3.

Figure 9. Four member stack with 2 Uplinks

FlexStack Comparison with StackWise Plus

The FlexStack feature is based on StackWise Plus. For those network administrators familiar with managing Cisco Catalyst 3750-E or 3750-X stacks running StackWise Plus, managing FlexStack is very similar. There are some differences between FlexStack and StackWise Plus. This section will describe the similarities as well as the differences.
Operationally FlexStack is similar to StackWise Plus. The similarities are in the way the switch is managed. There is a single management point of contact for the stack. All switch configuration is done from the master and applies globally. The stack appears to have a single forwarding plane for all data traffic. FlexStack supports high-availability cross-stack features such as EtherChannel, FlexLinks, and Spanning Tree Protocol.
The operational differences are minor and are related more to Cisco IOS Software feature sets on the two different families of Ethernet switches, rather than on operational aspects of managing switch stacks. The Cisco Catalyst 3750-E and 3750-X Series Switches use licensing to manage feature sets, whereas the Cisco Catalyst 2960-S switches do not have licenses for feature sets. The Cisco Catalyst 3750-E and 3750-X Series Switches support Layer 3 functionality (for example, Layer 3 forwarding and IP routing).
Table 2 shows the differences and the similarities for FlexStack and StackWise Plus.

Table 2. Comparing FlexStack and StackWise Plus


Cisco Catalyst 2960-S Stacking Ease of Use

Cisco Catalyst 3750-X StackWise Plus
Ease of Use and High Availability

Device limit

4 units

9 units

Stack bandwidth




Hardware drop table

Ring (destination stripping)

Dynamic ring load balancing



Stack convergence

1-2 seconds

Few milliseconds

Stack quality of service (QoS)

Applied hop by hop

Applied on ingress


Single IP address, SNMP, syslog

Single IP address, SNMP, syslog


Single config and CLI, auto image and config update

Single config and CLI, auto image and config update

Show and debug commands



Single forwarding and control plane

Synchronize Address Resolution Protocol (ARP), MAC address, Internet Group Management Protocol (IGMP), VLAN tables

Synchronize Address Resolution Protocol (ARP), MAC address, Internet Group Management Protocol (IGMP), VLAN tables, routing tables

Cross-stack features



Single bridge-ID




Stack master 1:N redundancy

Stack master 1:N redundancy

The differences between FlexStack and StackWise Plus are rooted in the different hardware architectures. StackWise Plus is the superior architecture. StackWise Plus is a dual redundant ring architecture with individual links operating at 32Gbps. The StackWise Plus connections are not pure Ethernet links but rather a ring architecture, allowing packets to take either ring. The FlexStack links are full duplex 10Gbps Ethernet links. The ring architecture gives StackWise Plus a significant advantage over FlexStack when it comes to high availability of the stack.
In Table 2 stack convergence is measured in milliseconds for StackWise Plus, as opposed to seconds for FlexStack. This is because of the ring architecture. In StackWise Plus packets ingressing the stack can be forwarded on either ring. When one ring becomes inoperable, all ingressing packets are forced to a single ring. The speed of one ring is 32Gbps, and can easily accommodate an entire stack. The recovery time for a single link going inoperable is not dependent upon the number of members in the stack. It is consistent across all stack sizes. In FlexStack, when a single link becomes inoperable, there is a stackwide disruption in the forwarding of packets. The members must rediscover the new topology. This rediscovery time can take up to 1-2 seconds depending upon the number of members in the stack. Smaller stacks will reconverge faster than larger stacks.
Because FlexStack stack links are Ethernet links, forwarding packets from one stack member to another requires the packets to be forwarded by each member. This is referred to as hop by hop, meaning the packets must hop from the ingressing member to the egressing member across all members in between. Not all members see the packet. Only those members in the data path will forward the packet. This hop-by-hop behavior is the same as different Ethernet switches forwarding the packet amongst themselves (at 10Gbps speeds). Hop by hop requires that each member must queue the packet. Queuing the packet means that QoS is applied at each hop. If a stack link between two members becomes congested, packets that successfully ingress on one member may be discarded by another member. There is no dynamic adjustment to the forwarding of packets when one link becomes congested. With StackWise Plus, once a packet is on the ring, it is guaranteed to be delivered to the destination interface. Add to this that StackWise Plus rings will dynamically load balance, and you can see that ring congestion is not an issue on StackWise Plus.

Troubleshooting Tips and Operational Details

FlexStack Member Addition

When a new switch joins an existing stack (that is, a stack that has already elected a master), then the new switch will be rebooted after it joins. This is to remove any old configuration. When the new switch completes the reload, it will become a member of the existing stack. The new member will be assigned the next available unused member number if it was not preprovisioned. Preprovisioning of switch members is covered in the Cisco Catalyst 2960-S software configuration guide on This behavior is covered in more detail in subsequent sections.
Details on FlexStack Cables and Connections
Match the keys on the cable with the connector when connecting FlexStack cables. It should slide in easily. When a good connection is made, there should be a "click" as the cable connects correctly. If it does not slide in easily, try turning the cable 180 degrees to match the keys on the cable with the connector. The tabs on the cables face away from each other.
Figure 10 shows a stack with a single bad FlexStack cable or connection (indicated by link B). This stack provides only half bandwidth and does not have redundant connections. A broken FlexStack link is operationally the same as not having a FlexStack link.

Figure 10. Failed Stack Link

To see if the stack has a broken or unconnected stack port, use the "show switch stack-ports" command. Stack port connections for the nonredundant stack shown in Figure 9:
Figure 11 shows a stack without redundant connections. When the FlexStack link indicated by B breaks, the FlexStack will break into two logical switches. The top two members will form one logical stack with two members, and the other logical stack will contain a single member.

Figure 11. Broken Stack

Since the stack in the connection in Figure 11 has broken into two, only two members will have their stack port status shown. The third member has been removed from the stack.
Stack port connections for the broken stack shown in Figure 11:


FlexStack technology provides network administrators the ease of use for managing multiple switches. Cisco Catalyst 2960-S switches are the fully featured Layer 2 switches that use FlexStack. FlexStack has built-in hardware redundancy along with built-in ease of use.