Struct CheckpointManager 

pub struct CheckpointManager<S: CheckpointStore = InMemoryCheckpointStore> { /* private fields */ }

Drives periodic checkpointing and provides recovery support.

Call Self::on_event after processing each event. When the cumulative sequence number reaches the next scheduled checkpoint, a new CheckpointState is automatically saved to the underlying store.

Implementations

impl<S: CheckpointStore> CheckpointManager<S>

pub fn new(store: S, checkpoint_interval: u64) -> Self

Create a manager with the given store and interval.

Use a durable store such as FileCheckpointStore for recovery that survives a process restart, or InMemoryCheckpointStore for tests and single-process, non-recoverable use.

pub fn on_event( &mut self, stream_id: &str, sequence: u64, watermark_ns: u64, ) -> Result<bool, StreamingError>

Called after each processed event.

Returns Ok(true) if a checkpoint was taken, Ok(false) otherwise.

pub fn recover(&self, stream_id: &str) -> Result<Option<u64>, StreamingError>

Return the sequence number from which to resume, or None if no checkpoint exists for the stream.

With a durable store this reads the last checkpoint persisted before a crash, enabling minimal-replay recovery.

pub fn total_checkpoints(&self) -> u64

Total checkpoints taken since this manager was created.

pub fn store(&self) -> &S

Read-only access to the underlying store (for inspection / testing).