Struct ExpirationService 

pub struct ExpirationService {
    index: Arc<dyn IndexBackend + Send + Sync>,
    config: ExpirationConfig,
}

Service for garbage collecting memories that have exceeded their TTL.

Unlike RetentionGarbageCollector which uses retention policies based on age, this service checks the explicit expires_at timestamp set at capture time.

Thread Safety

The service holds an Arc reference to the index backend, making it safe to share across threads.

Fields

index: Arc<dyn IndexBackend + Send + Sync>

Reference to the index backend for querying and updating memories.

config: ExpirationConfig

Expiration configuration.

Implementations

impl ExpirationService

pub fn new( index: Arc<dyn IndexBackend + Send + Sync>, config: ExpirationConfig, ) -> Self

Creates a new expiration service.

Arguments

• index - Shared reference to the index backend.
• config - Expiration configuration.

pub fn should_run_cleanup(&self) -> bool

Determines whether cleanup should run based on configured probability.

Uses a random number generator to decide probabilistically. This enables lazy, opportunistic cleanup without expensive full scans on every operation.

Returns

true if cleanup should run, false otherwise.

pub fn gc_expired_memories(&self, dry_run: bool) -> Result<ExpirationGcResult>

Performs garbage collection on TTL-expired memories.

This method:

1. Lists all active (non-tombstoned) memories
2. Checks each memory for TTL expiration (expires_at < now)
3. Tombstones expired memories (unless dry_run)

Arguments

• dry_run - If true, only report what would be done without making changes

Returns

An ExpirationGcResult containing statistics about the operation.

Errors

Returns an error if index backend operations fail.

pub const fn config(&self) -> &ExpirationConfig

Returns the current expiration configuration.