Table of Contents
This document introduces the Access Point Stateful Switch Over redundancy model for High Availability (HA) with CT5760 controllers using the StackWise-480 technology. HA in Cisco 5700 Series Wireless Controller is enabled using Cisco StackWise-480 technology. This deployment guide applies only to the Converged Access CT5760 Wireless LAN Controller. For more information on the CT5760 please refer to CT5760 Controller Deployment Guide.
CT5760 is an innovative UADP ASIC based wireless controller deployed as a centralized controller in the next generation unified wireless architecture. CT5760 controllers are specifically designed to function as Unified model central wireless controllers. They also support the newer Mobility functionality with Converged Access switches in the wireless architecture.
As a component of the Cisco Unified Wireless Network, the CT5760 series works in conjunction with Cisco Aironet access points, the Cisco Prime infrastructure, and the Cisco Mobility Services Engine to support business-critical wireless data, voice, and video applications.
IOS XE 3.3 SE release for Cisco 5700 Series Wireless Controller introduces 1:1 Active-Standby redundancy model for HA with the CT5760 controllers using the StackWise-480 technology. HA in Cisco 5700 Series Wireless Controller is enabled using Cisco StackWise-480 technology. StackWise-480 identifies active and standby members in the stack as per the Cisco IOS Software SSO technology. All the control plane activities are centralized and synchronized between the active and standby units. The Active Controller centrally manages all the control and management communication. The network control data traffic is transparently switched from the standby unit to the active unit for centralized processing. The 12 ports of the CT5760 HA redundant pair can be connected to the infrastructure network either through a single switch or two switches.
Bulk and Incremental configuration is synced between the two controllers at run-time and both controllers share the same IP address on the management interface. The CAPWAP state of the Access Points that are in Run State is also synched from the active WLC to the Hot-Standby WLC allowing the Access Points to be state-fully switched over when the Active WLC fails. The APs do not go to the Discovery state when Active WLC fails, and Standby WLC takes over as the Active WLC to serve the network.
There is no preempt functionality with SSO meaning that when the previous Active WLC resumes operation, it will not take back the role as an Active WLC but will negotiate its state with the current Active WLC and transition to the Hot-Standby state. The Active and Standby election process is an automated election process that occurs as part of the SSO StackWise technology and will be explained later in the document.
HA with AP SSO is supported in Cisco 5700 Series Wireless Controllers. IOS XE 3.3 release only supports AP SSO i.e. APs will not disconnect and continue to be associated to the controller after a switchover. However, all clients will be de-authenticated and forced to rejoin new the Active WLC because Client SSO is not supported with this release.
A CT5760 HA Pair is a special case of a switch stack that can have up to two CT5760 controllers connected through their StackWise-480 ports. The stack members work together as a unified system. A third CT5760 cannot join the switch stack or HA pair. A switch stack always has one active controller and one standby controller. If the active controller becomes unavailable, the standby assumes the role of the active, and continues to the keep the stack operational. The active controller controls the operation of the HA pair, and is the single point of stack-wide management. The term switch is loosely used in the document to refer to the CT5760 WLC for this reason.
StackWise-480 has a stack bandwidth of 480 Gbps and uses SSO to provide resiliency within the HA Pair. The Active CT5760 WLC creates and updates all the wireless information and constantly synchronizes that information with the standby controller. If the active WLC fails, the standby WLC assumes the role of the active WLC and continues to the keep the HA Pair operational. Access Points continue to remain connected during an active-to-standby switchover.
- Port Status is OK i.e. The cable is detected and connected neighbor is up.
- Link OK is Yes i.e. The stack cable connected.
- Link Active is Yes i.e. The neighbor is detected. The Port can send traffic over this link.
By default, the Local MAC Address of the HA Pair is the MAC address of the current Active Controller. When a switchover occurs, the MAC address changes to the MAC address of the new Active Controller.
- Enter the command without a value or with a value of 0 to continue using the MAC Address of the current active controller indefinitely.
- Enter a time-value from 1 to 60 minutes to configure the time period before the redundant unit MAC Address changes to the new Active Controller. The MAC address of the previous active controller is used until the configured time period expires.
The active controller can be accessed through a console connection, Telnet, an SSH, or a Web Browser by using the Management IP address. To access the IOS console of the standby WLC, use the
session standby iosprivileged EXEC command from the active CT5760 after executing the following commands from the config mode.
- The CT5760 WLC HA pair is reset.
- The active CT5760 WLC is removed from the HA pair.
- The active CT5760 WLC is reset or powered off.
- The active CT5760 WLC fails.
- Powered-on standalone CT5760 WLC is merged with the current Active CT5760 WLC.
Note We recommend assigning the highest priority value to the WLC 5760 you prefer to be the active controller. This ensures that the controller is re-elected as active controller if a re-election occurs.
You can display the current priority value by using the
show switchuser EXEC command. The new priority value takes effect immediately but does not affect the current Active Controller. The new priority value helps determine which controller is elected as the new Active Controller when the current active CT5760 WLC or HA redundant pair reloads.
- Adding a powered-on CT5760 WLC (merging) to an existing Active Controller causes both WLCs to reload and elect a new active controller from among themselves.
- Make sure that the controller to be paired is powered down before connecting it to the existing Active Controller by using Stack Cables. Once the Stack Cables are connected, power on the new controller. The newly introduced controller will take up its role as a Hot-Standby controller because an Active Controller already exists.
- Similarly, removing a powered-on WLC causes both WLCs to reload. Power down the controller that needs to be removed before disconnecting the Stack Cables to avoid this.
The LAG, or an EtherChannel, bundles all the existing ports in both the standby and active units into a single logical port to provide an aggregate bandwidth of 60 Gbps. The creation of an EtherChannel enables protection against port failures and provides port level redundancy. The EtherChannels or LAGs created are used for link redundancy to ensure HA of Access Points. Multiple LAGs can be configured on the CT5760 WLC.
The offline configuration feature can be used to provision (to supply a configuration to) a new CT5760 WLC before it is paired with the Active Controller. You can configure the stack member number, the switch type, and the interfaces associated with a WLC that is not currently part of the HA redundant pair. The configuration that is created on the active WLC is called the provisioned configuration. The WLC that is paired with the active controller and that receives this configuration is called the provisioned switch.
The provisioned configuration is manually created through the
provision typeglobal configuration command. The stack-member-number and type on the provisioned CT5760 must match the stack member number and type that was created for the new controller before it is paired with the active controller.
When you configure the interfaces associated with a provisioned CT5760 WLC, the active controller accepts the configuration and the information appears in the running configuration. However, because the second controller is not active, any configuration on the interface is not operational and the interface associated with the provisioned unit does not appear in the display of the specific feature. For example, VLAN configuration information associated with a provisioned WLC does not appear in the
show vlanuser EXEC command output. The provisioned configuration is retained in the running configuration whether or not the provisioned WLC is part of the HA pair. You can save the provisioned configuration to the startup configuration file by entering the
copy running-config startup-configprivileged EXEC command.
An Access Point Stateful Switch Over (AP SSO) implies that all the Access Point sessions are switched over state-fully and Access Points continue to operate in a network with no loss of sessions, providing improved network availability and reducing service downtime. The Active Controller in the redundant pair is equipped to support 1000 Access Points and 12000 clients.
show redundancydisplays the redundant system and the current processor information. The redundant system information includes the system uptime, standby failures, switchover reason, hardware mode, and configured and operating redundancy mode. The current processor information displayed includes the image version, active location, software state, BOOT variable, configuration register value, and uptime in the current state, and so on. The Peer Processor information is only available from the Active Controller.
AP CAPWAP state information is synced to the standby controller. AP uptime and association time can be verified using the command
show ap summaryand
show ap uptimeon the Active and Standby Controllers.
The 12 ports of CT5760 WLC can be connected to the infrastructure network either through a single switch or through two switches. In either case, the ports of CT5760 WLC are configured into LAG bundles. Failing of ports connected to neighbor switches does not induce a system switchover and it only affects bandwidth when less than 6 ports are active. However, system switchover can occur due to various reasons identified below.
This type of switch over occurs when any of the key processes running on the Active unit fails or crashes. Examples of this are HA Manager, WCM, Stack Manager, FED, FFM, and so on. Upon such a failure, the Active unit reloads and the hot Standby takes over and becomes the new Active unit. When the failed system boots up, it will transition to Hot-Standby state. If the Standby unit is not yet in Hot Standby State, both units are reloaded and there will be no AP SSO. A process failure on the standby (hot or not) will cause it to reload.
This switchover from the Active to Standby unit is caused due to power failure of the current Active unit. The current Standby unit becomes the new Active unit and when the failed system boots up, it will transition to Hot-Standby state.
The HA feature is enabled by default when CT5760 WLCs are connected using the stack cable and the Cisco StackWise-480 technology is enabled. You cannot disable it; however, you can initiate a manual graceful switchover using the command line interface. This is a user initiated forced switchover between the Active and Standby unit. The current Standby unit becomes the new Active unit and when the failed system boots up, it will transition to Hot-Standby state. To perform a manual switchover, execute the
redundancy force-switchovercommand. This command initiates a graceful switchover from the active to the standby controller. The active controller reloads and the standby takes over as the New Active controller.
Prior to IOS XE Release 3.3, AP SSO was not supported on CT5760 WLCs. Upgrade individual controllers to image version 3.3 before pairing them for HA using the guidelines explained in the Best Practices for HA Pairing Process section.
A hybrid topology of AP SSO redundant pair and N+1 primary, secondary and tertiary model is supported as shown above. The secondary controller at the DR site can be a CT5760 or a CUWN controller running New Mobility. Access points failing back from converged access CT5760 WLC to CUWN controllers will re-download the code before joining the CUWN WLC.
In a HA redundant pair, the AP count is combined from license counts on both controllers. The Active controller will display all AP licenses broken down per controller. The total number of APs that can be supported by the CT5760 WLC HA pair is 1000.
When CT5760 WLCs are paired together to form a HA pair, the cumulative AP count may exceed 1000. However, the WCM module on the controller does the gate keeping of not allowing more than 1000 APs on the redundant HA pair. A warning message on the console indicates to the user that the cumulative AP count has exceeded the maximum AP count of 1000.
In the remove case, the in-use AP count may exceed available AP count licenses on the active controller. If the total number of APs connected exceeds the available AP count in the HA pair, syslog and warning messages are displayed that those APs would be disconnected on AP or Stack reboot. The above can happen if a deleted controller takes away the AP count with it or by configuration, where the customer reduces number of adder licenses. In general, when a controller is removed from the HA pair, it takes the previously installed AP count with it. However, the active controller will continue to support the AP count of that lost controller in order not to interrupt the service, but upon either a controller or an AP reboot, some APs will not be able to join back and the APs that can join the controller will fall back to the AP count available on the active controller, i.e., the AP count licenses originally on that controller.
- The 5760 HA-SKU provides support for up to 1000 APs when Active controller Fails over to HA-SKU unit.
- The following notification is generated on the HA-SKU controller when a Standby non HA-SKU is lost:
- Two 5760 HA-SKU units cannot be connected to be paired in an SSO pair.
- Licenses cannot be added to a 5760 HA-SKU unit.
- The 5760 HA-SKU has a PID AIR-CT5760-HA-K9 to indicate that it is a HA-SKU unit.
- A non HA-SKU 5760 controller cannot be converted to a HA-SKU by configuration.
- Switchover during AP Pre-Image download causes the APs to start image download all over again from the new Active controller.
- Rogue APs and clients are not synced to Standby and are re-learnt upon switchover.
- Infrastructure MFP key is not synced to the Standby controller and is re-learnt upon switchover.
- New Active controller re-learns the shun list from IPS and other MCs, and redistributes it to the MAs.
- wIPS information is not synced to the Standby unit and is re-learnt upon switchover.
- Clean Air detected Interferer devices are re-learnt after switchover.
- Net Flow records are cleared upon switchover and collection starts fresh on the new Active controller.
- Mobility paths and tunnels to the MO and other peer MCs are not disrupted upon switchover. However the Client state is cleaned up on the MO under which the HA pair exists and is re-learnt from the new Active controller when the client re-associates.
- Roamed clients that have their data path going through the Mobility Tunnel Endpoint (MTE) “become Local” in case of L2 with Sticky Anchoring and L3 Roam. L2 Roamed Clients are not affected except when roaming occurs between CUWN and CA controllers.
- RRM related configurations and the AP neighbor list in the Leader HA pair is synced to the Standby controller.
- Upon Guest Anchor controller switchover, mobility tunnels stay active, APs remain connected, clients rejoin at MA or MC, and are anchored on the new Active controller.