Struct LeaseLifecycle 

pub struct LeaseLifecycle { /* private fields */ }

Drives the serverless writer lease for one database key.

Owns the WriterLease snapshot, the LeaseStore it talks to, and the side-effect channels (WriteGate, AuditLogger, MarkDraining). All transitions go through methods on this struct so the gate/audit pair stays consistent.

Implementations

impl LeaseLifecycle

pub fn new( store: Arc<LeaseStore>, write_gate: Arc<WriteGate>, audit_log: Arc<AuditLogger>, mark_draining: MarkDraining, holder_id: String, database_key: String, ttl_ms: u64, ) -> Self

pub fn holder_id(&self) -> &str

pub fn database_key(&self) -> &str

pub fn ttl_ms(&self) -> u64

pub fn current_lease(&self) -> Option<WriterLease>

pub fn try_acquire(&self) -> RedDBResult<()>

Acquire the writer lease and flip the gate to Held. Audit line records the outcome (ok / err) either way.

pub fn refresh(&self) -> RedDBResult<()>

Refresh the held lease. On success, updates the in-memory snapshot and returns Ok(()). On any backend error, flips the gate to NotHeld, audits lease/lost, asks the runtime to drain, and returns Err.

The refresh thread should treat Err as terminal and exit — re-acquiring after a loss could race a freshly promoted writer.

pub fn release(&self) -> RedDBResult<()>

Release the held lease and flip the gate to NotHeld. Best- effort: a backend failure is logged but does not propagate — shutdown should not block on a remote release.