Struct CheckpointManager 

pub struct CheckpointManager { /* private fields */ }

Manages periodic checkpointing to bound recovery time.

Checkpointing ensures that:

1. All data up to a certain LSN is persisted to the main database file
2. WAL segments before that LSN can be truncated
3. Recovery only needs to replay WAL records after the checkpoint

Checkpoint Process

1. Record current WAL LSN as checkpoint_lsn
2. Persist all in-memory data to the main database file
3. fsync the main database file
4. Update meta page with checkpoint_lsn
5. fsync meta page
6. Truncate WAL segments before checkpoint_lsn

Implementations

impl CheckpointManager

pub fn new(config: CheckpointConfig) -> Self

Creates a new checkpoint manager with the given configuration.

pub fn restore( config: CheckpointConfig, checkpoint_info: CheckpointInfo, ) -> Self

Restores checkpoint manager state from persisted checkpoint info.

pub fn should_checkpoint(&self, wal: &Wal) -> bool

Determines if a checkpoint should be performed based on configured thresholds.

A checkpoint is triggered if any of these conditions are met:

• Time since last checkpoint exceeds config.interval
• WAL growth since checkpoint exceeds config.wal_threshold
• Records since checkpoint exceed config.min_records

pub fn record_writes(&mut self, count: usize)

Records that records have been written since last checkpoint.

pub fn record_checkpoint(&mut self, lsn: Lsn)

Records that a checkpoint was completed at the given LSN.

pub fn record_checkpoint_with_wal_size(&mut self, lsn: Lsn, wal_size: u64)

Records that a checkpoint was completed at the given LSN, tracking WAL size.

pub fn last_checkpoint_lsn(&self) -> Lsn

Returns the LSN of the last completed checkpoint.

pub fn create_checkpoint_info( &self, lsn: Lsn, entry_count: u64, ) -> CheckpointInfo

Creates checkpoint info for the given LSN.