Struct ReconnectingClient 

pub struct ReconnectingClient { /* private fields */ }

Reconnecting client wrapper offering automatic reconnection and leader redirection support per Architecture §21.

Implementations

impl ReconnectingClient

pub async fn connect(addr: &str) -> Result<Self, ClientError>

Create a new reconnecting client

The address can be either an IP:port (e.g., “127.0.0.1:1992”) or a hostname:port (e.g., “lance.example.com:1992”).

pub fn from_existing(client: LanceClient, addr: &str) -> Self

Wrap an existing LanceClient with auto-reconnect support.

The addr is used for DNS re-resolution on reconnect (important for load-balanced endpoints). By default, retries are unlimited.

pub async fn connect_tls( addr: &str, tls_config: TlsClientConfig, ) -> Result<Self, ClientError>

Create a new reconnecting client with TLS

The address can be either an IP:port (e.g., “127.0.0.1:1992”) or a hostname:port (e.g., “lance.example.com:1992”).

pub fn with_max_attempts(self, attempts: u32) -> Self

Set maximum reconnection attempts (0 = unlimited)

pub fn with_unlimited_retries(self) -> Self

Configure for unlimited reconnection attempts (never give up)

pub fn with_base_delay(self, delay: Duration) -> Self

Set base delay for exponential backoff

pub fn with_max_delay(self, delay: Duration) -> Self

Set maximum delay for exponential backoff

pub fn with_follow_leader(self, follow: bool) -> Self

Enable or disable automatic leader following

pub fn original_addr(&self) -> &str

Get the original connection address

pub fn leader_addr(&self) -> Option<SocketAddr>

Get the current leader address if known

pub fn set_leader_addr(&mut self, addr: SocketAddr)

Update the known leader address (called when redirect received)

pub fn reconnect_attempts(&self) -> u32

Get total reconnection attempts made

pub async fn client(&mut self) -> Result<&mut LanceClient, ClientError>

Get a reference to the underlying client, reconnecting if needed

pub async fn reconnect(&mut self) -> Result<(), ClientError>

Attempt to reconnect with exponential backoff and DNS re-resolution. On success, resets the reconnect attempt counter.

pub async fn execute<F, T>(&mut self, op: F) -> Result<T, ClientError>

where F: Fn(&mut LanceClient) -> Pin<Box<dyn Future<Output = Result<T, ClientError>> + Send + '_>>,

Execute an operation with automatic reconnection on failure. Retries on all retryable errors (connection failures, FORWARD_FAILED, timeouts, backpressure, NOT_LEADER) with exponential backoff.

pub fn mark_failed(&mut self)

Mark connection as failed