Struct PeerConnection 

pub struct PeerConnection {
    pub config: PeerConfig,
    /* private fields */
}

A managed connection to a peer’s Redis.

Uses redis::aio::ConnectionManager internally, which provides:

• Automatic reconnection on connection loss
• Multiplexed connection (single TCP socket, multiple in-flight requests)
• Connection pooling semantics (cloning is cheap, shares underlying connection)

Fields

config: PeerConfig

Peer configuration

Implementations

impl PeerConnection

pub fn new(config: PeerConfig) -> Self

Create a new peer connection (not yet connected).

pub fn node_id(&self) -> &str

Get the peer’s node ID.

pub fn cdc_stream_key(&self) -> String

Get the CDC stream key for this peer.

pub async fn state(&self) -> PeerState

Get current connection state.

pub async fn is_connected(&self) -> bool

Check if connected.

pub async fn circuit_state(&self) -> PeerCircuitState

Get the current circuit breaker state.

pub async fn is_circuit_open(&self) -> bool

Check if the circuit is open (should reject requests).

pub async fn record_success(&self)

Record a successful operation (resets failure count).

pub async fn record_failure(&self)

Record a failed operation (increments failure count, may open circuit).

pub async fn connect(&self, retry_config: &RetryConfig) -> Result<()>

Connect to the peer’s Redis with retry logic.

pub async fn connection(&self) -> Option<ConnectionManager>

Get the stream connection (for XREAD and other potentially blocking ops).

Returns None if not connected.

pub async fn query_connection(&self) -> Option<ConnectionManager>

Get the query connection (for GET, HGET, and other fast point queries).

This is a separate connection that won’t be blocked by XREAD BLOCK on the stream connection.

Returns None if not connected.

pub async fn ensure_connected(&self) -> Result<ConnectionManager>

Ensure the peer is connected, connecting lazily if needed.

This provides a convenient way to get a connection without manually calling connect() first. Uses default retry config.

Returns

The connection manager, or an error if connection failed.

pub fn failure_count(&self) -> u64

Get consecutive failure count.

pub fn millis_since_success(&self) -> u64

Get milliseconds since last success.

pub async fn mark_disconnected(&self)

Mark connection as failed (triggers reconnect).

pub fn shutdown(&self)

Signal shutdown.

pub async fn ping(&self) -> Result<Duration>

Ping the peer’s Redis to check connection health.

Returns the round-trip latency on success. Updates last_success timestamp on success.

pub async fn get_merkle_root(&self) -> Result<Option<[u8; 32]>>

Get the peer’s Merkle root hash (cached with TTL).

Uses a short-lived cache to avoid hammering the peer during repair cycles. The cache is automatically invalidated after MERKLE_CACHE_TTL.

pub async fn invalidate_merkle_cache(&self)

Invalidate the Merkle root cache (call after local writes).

pub async fn get_merkle_children( &self, path: &str, ) -> Result<Vec<(String, [u8; 32])>>

Get children of a Merkle path (sorted by score/position).

pub async fn get_item(&self, key: &str) -> Result<Option<Vec<u8>>>

Get an item’s data by key.

Returns the raw item content from the peer’s Redis. Uses JSON.GET for JSON items, GET for binary items.