Monitor Global-Level Objects in a Chassis Cluster

SPU Monitoring

Services Processing Units (SPUs) monitoring tracks the health of the SPU and the central point (CP). The chassis manager on each SPC monitors the SPUs and the central point, and maintains a heartbeat with the Routing Engine process, chassisd. In this hierarchical monitoring architecture, chassisd serves as the central component for hardware failure detection. SPU monitoring is enabled by default.

A persistent failure of the SPU and central point on a node is treated as a catastrophic Packet Forwarding Engine (PFE) failure. In this scenario, the node's PFE is disabled within the cluster by reducing the priorities of redundancy groups x to 0.

• A central point failure triggers a failover to the secondary node. The failed node's PFE, including all SPCs and Input/Output cards (IOCs), is automatically restarted. If the secondary central point has also failed, the cluster cannot come up because no primary device is available. In this scenario, only the data plane (redundancy group x) fails over.

• A single SPU failure triggers a failover of redundancy group x to the secondary node. All IOCs and SPCs on the failed node are automatically restarted, and redundancy group x fails over to the secondary node without requiring user intervention. When the failed (former primary) node recovers, failback behavior is determined by the preempt configuration for redundancy group x.

  The dead SPU detection interval is 30 seconds.

This event triggers an alarm indicating that a new field-replaceable unit (FRU) is required.

Flowd Monitoring

Flowd monitoring tracks the health of the flowd process and is enabled by default.

A persistent failure of the flowd process on a node is considered a catastrophic Packet Forwarding Engine (PFE) failure. In this case, the node's PFE is disabled in the cluster by reducing the priorities of redundancy groups x to 0.

A flowd process failure triggers an automatic failover of redundancy group x to the secondary node without requiring user intervention. When the failed (former primary) node recovers, failback behavior is determined by the preempt configuration for redundancy group x.

During SPC or flowd monitoring failures on a local node, the data plane redundancy group RG1+ fails over to the peer node that is in a healthy state. However, the control plane redundancy group RG0 does not fail over and remains primary on the same node as before the failure.

Cold-Sync Monitoring

The process of synchronizing data plane runtime objects (RTOs) when SPUs or the flowd process start is known as cold synchronization cold sync. Cold sync ensures that all RTOs are fully synchronized between nodes before an SPU or flowd instance is eligible for the primary role. Once all RTOs are synchronized, the cold-sync process is complete and the SPU or flowd on the node is ready to take over, if required.

Cold-sync monitoring tracks the completion status of cold synchronization across all SPUs or flowd instances on a node and is enabled by default. When preemption is enabled, cold-sync monitoring prevents a node from assuming the primary role until cold synchronization has completed successfully for all SPUs or flowd processes on that node.

When a node reboots, or when SPUs or flowd recover from a failure, the priorities of all redundancy groups RG1+ are set to 0. At each SPU or flowd instance comes online, it initiates cold-synchronization with its corresponding peer on the other node.

If the node is operating as the only node in the cluster, the priorities of all RG1+ groups remain at 0 until a second node joins the cluster. While operating at priority 0, the device can continue to send and receive traffic, but it cannot participate in failover.

When a new node joins the cluster, SPUs or flowd instances on both nodes initiate cold synchronization. Upon detecting a cold-sync request, the SPU or flowd on the existing node signals completion. The SPUs or flowd on the newly joined node signal completion only after all RTOs have been fully synchronized.

After completion messages are received from all SPUs or flowd processes, and if no other monitored component failures exist (such as interface failures), the priorities of RG1+ transition from 0 to their configured values. This mechanism ensures that the existing primary node regains its configured priority first, while the newly joined node transitions to its configured priority only after completing cold synchronization. As a result, the newly added node is fully synchronized and ready before it can assume the primary role.

Platform-Specific Monitor Objects Behavior

Use Feature Explorer to confirm platform and release support for specific features.

Use the following table to review platform-specific behaviors on your platform.