Consensus State

Overview

Consensus state records where a node is within the agreement process. It can include the current height, round, validator set, locked proposal, received votes, justified or finalized checkpoints, randomness, and other variables required by the protocol state machine.

This state differs from application state such as account balances and contract storage. Consensus state determines which execution result is canonical, while application execution computes the contents committed by that result. Node software must persist and restore safety-critical fields without double voting or accepting stale context. Operators should back up only supported data, protect validator signing history, and verify state after recovery. Applications generally consume finalized outputs rather than modifying consensus state directly. Consensus State should be tied to a specific network, block hash, protocol version, and state-transition rules. For Consensus State, a state value without that context can become ambiguous after a reorganization, upgrade, or replay on another network even when the serialized bytes appear valid.

A commitment associated with Consensus State proves only what its verification rules establish. For Consensus State, commitment integrity, data availability, execution correctness, and historical retrievability are separate properties; a valid root or proof does not by itself guarantee that every application can reconstruct the underlying state. Archive requirements, pruning, snapshots, and state expiry can change which historical observations remain available from ordinary infrastructure. Applications should record the canonical block reference, query method, proof or response, and node capability used to obtain it. Implementations of Consensus State need deterministic encoding and version-aware validation. For Consensus State, cross-client tests should cover empty values, boundary cases, upgraded data structures, and reorganization recovery so the same evidence produces the same interpretation across independent systems.