Requirements For Teaming - HP HPE VAN SDN Controller 2.7 Administrator's Manual

Table of Contents

For the controllers in a team to remain active, they must be part of the team quorum. To be

part of a team quorum, a controller must be connected to at least one other team member

that has a status of active or initializing. If one controller in the team goes offline, controller

operations can continue. However, if two controllers in a team fail, the third controller does

not operate as a standalone controller. Instead, the third controller loses its membership in

the team quorum, and the controller status is changed to suspended.

A region groups devices together with their controllers. A region must have three controllers

which must be specified in priority order for all devices within the region (master, primary

slave, secondary slave).

Putting the region configuration in place for a controller team ensures seamless failover and

failback among the configured controllers for the specified network devices in a region. When

a controller experiences a fault, the region configuration ensures that a slave controller

immediately assumes the master role over the groups of network devices for which the failed

controller was master. Once the failed controller recovers and rejoins the team, the rejoining

controller takes back the role for which it was configured with respect to the network devices.

For details on failover and failback, see

"Failback behavior within a region" (page

Devices in a region can be expressed as a list of individual IPv4 addresses, a list of IPv4

ranges, or a combination of both. Devices included in a region can connect to the region's

controllers. OpenFlow 1.3 devices must be configured with the IPs of all three controllers in

a team. This allows one of those controllers to assert itself as the master of a given device.

The device then automatically assigns a role of slave to the other two configured controllers.

This ensures the master knows of all the events happening on the device while the slaves

are kept up to date on a subset of events.

Applications are stopped when there is a change in the teaming status for a given controller.

For example:

◦

When the controller transitions from stand-alone to a member of a team, all applications

are stopped prior to the creation of the team, and then restarted after the team is formed.

◦

When a teamed controller detects that it is no longer part of the quorum, all applications

are stopped until the controller detects it has rejoined the quorum.

All team members in a teamed environment must be active before you can make configuration,

licensing, application or region changes. Changes attempted when a team member is initializing

or disconnected are not guaranteed to be consistent, such changes are blocked through the

REST API and the GUI. Changes attempted through the REST API result in Forbidden response

(error code 403) and an UnsafeConfigurationException exception.

The team status can be: active, unreachable, or unknown (for details, see

(page

101)). You can view team status from the Team screen in the controller UI, see

team status" (page

The controller status can be: initializing, active, suspended, or unreachable (for details, see

"Controller status " (page

UI, see

"Viewing team configuration and controller status" (page

You can view region configuration from the Team screen in the controller UI, see

configuration" (page