Owner's Manual

Checkpoints

Data originating from configurations or dynamic operations in the active controller is included in the

checkpoint data.

• Configuration through REST APIs — The data resides on the active controller and in the database. The

standby controller reads data from the database to build its own local data copy as part of the role

transition process.

• Dynamic — The data is computed by the controller or learned through OF communication with

switches.

Checkpoint data is dynamically communicated by the active controller to the standby controller. This

process involves the following actions:

• The active controller provides a snapshot of the dynamically learned and computed data to the

standby controller.

• The HA framework notifies the other controller software components to send snapshot data after a

connection with the standby controller is established.

• The active controller performs event-based synchronization to update the standby controller.

When the standby controller starts up, the following process completes:

1. During initialization, the standby controller notifies the active controller that it is available.

2. The standby controller waits for a snapshot of the dynamic data from the primary controller.

3. The active controller prepares the snapshot and sends it to the standby controller.

NOTE: Do not modify the data while the snapshot is being prepared.

4. The standby controller processes the received snapshot.

When the controller is determined to be active, it restores the data from the database. After role

assignment, the standby controller waits for snapshot data from the active controller. The active-standby

synchronization process is initiated by the standby controller. After this initial synchronization, the rest of

the process is event-driven and updates the standby controller based on updates from the active

controller.

1. The active controller recognizes the standby controller when it receives the Standby_Is_Ready

message from the standby controller. The cluster coordinator server establishes the relationship.

After the standby controller is detected, the active controller creates a high-level snapshot, which

includes topologies, SMAC learn values, and VLT configurations.

2. The HA framework sends a snapshot request to all network entities registered with the HA manager.

After receiving this notification, each network element prepares its snapshot data and sends it to the

standby controller.

3. The HA manager provides the infrastructure to transfer the data. Each network element opens its

own connection to receive the snapshot request and send the snapshot data.

4. The HA framework uses the thrift mechanism to manage data serialization and deserialization.

NOTE: During the Snapshot in Progress state, the active controller does not process other

events that would change the snapshot data. Instead, the events are queued for further

processing after the snapshot process completes.

Reconciliation Process

When an active controller is unavailable, the standby controller detects this and requests a role change to

Active. The controller notifies the switches of the role change and begins the reconciliation process with

Supported Technologies and Protocols