Struct ShardEngine 

pub struct ShardEngine { /* private fields */ }

The shard engine coordinates N worker threads, each owning a ShardStore.

Commands are routed to the appropriate worker based on key hash.

Implementations

impl ShardEngine

pub fn new(shard_count: usize) -> Self

Create a new ShardEngine with the given number of worker threads.

pub fn new_with_storage( shard_count: usize, wal_writers: Vec<Option<Box<dyn WalWriter>>>, ) -> Self

Create a new ShardEngine with per-shard WAL writers.

Each element in wal_writers corresponds to one shard. The WAL writer is moved into the shard’s worker thread and called after every mutation.

pub fn new_with_recovery( shard_count: usize, wal_writers: Vec<Option<Box<dyn WalWriter>>>, recovery_fns: Option<Vec<Box<dyn FnOnce(usize, &mut ShardStore) + Send>>>, ) -> Self

Create a new ShardEngine with per-shard WAL writers and optional per-shard recovery callbacks.

Each recovery callback receives the shard index and a mutable reference to the freshly created ShardStore, allowing callers to replay RDB snapshots and WAL entries before the worker starts accepting commands.

pub fn shard_count(&self) -> usize

Get the number of shards.

pub fn dispatch(&self, cmd: Command) -> ResponseReceiver

Dispatch a command and return a receiver for the response.

pub fn dispatch_blocking(&self, cmd: Command) -> CommandResponse

Dispatch a command and block until the response is received.

pub fn dispatch_batch_blocking( &self, commands: Vec<Command>, ) -> Vec<CommandResponse>

Dispatch a batch of commands and block until all responses are received.

Commands that target a single key are grouped per shard and executed in-order, reducing response-channel overhead for pipelined workloads.

Trait Implementations

impl Drop for ShardEngine

fn drop(&mut self)

Executes the destructor for this type.