Module replication

Expand description

Replication Module

Implements single-primary, multi-replica replication via WAL streaming.

§Architecture

Primary: accepts writes and streams WAL records to replicas
Replica: read-only, connects to primary for WAL streaming
Initial sync via snapshot transfer, then incremental WAL

§Usage

// Primary
let options = RedDBOptions::persistent("./primary-data")
    .with_replication(ReplicationConfig::primary());

// Replica
let options = RedDBOptions::persistent("./replica-data")
    .with_replication(ReplicationConfig::replica("http://primary:50051"));

Modules§

bookmark: Compatibility reexport for causal bookmark wire tokens.
cascade: Cascading replication for async read-replicas (issue #838, PRD #819).
cdc: Change Data Capture (CDC) — stream of database change events.
commit_policy: Primary commit policies (PLAN.md Phase 11.4).
commit_waiter: Synchronous commit waiter (PLAN.md Phase 11.4 — ack_n).
control_plane: Control-plane consensus seam (issue #996, parent #987, ADR 0052).
election: Term-based, quorum-gated automatic election (issue #834, PRD #819, ADR 0030).
failover: Coordinated zero-RPO failover (issue #833, PRD #819).
fence: Stale-term fencing for a returning ex-primary (issue #835, PRD #819, ADR 0030).
flow_control: Write-admission flow control keyed on in-quorum replica lag (issue #826).
lease: Serverless writer lease (PLAN.md Phase 5 / W6).
logical: Logical replication helpers shared by replica apply and point-in-time restore.
primary: Primary-side replication: WAL record production and snapshot serving.
quorum: Quorum-based commit coordination (Phase 2.6 multi-region PG parity).
replica: Replica-side replication: connects to primary, consumes WAL records.
rollback: Auto-rollback of a deposed primary to the common point (issue #840, PRD #819, ADR 0030).
scheduler: Backup Scheduler — automatic periodic snapshots with optional remote upload.
swap_db: Stay-readable re-bootstrap with an atomic dataset swap (issue #837, PRD #819).
topology_advertiser: Server-side TopologyAdvertiser (issue #167).
witness: Witness runtime profile (issue #836, PRD #819, ADR 0030).

Structs§

CascadeRelay: Tracks the sub-replica slots an intermediate holds and the frontiers that must propagate through the chain. Pure bookkeeping — the forwarding transport calls into it to decide what to send and what to advertise upstream.
CascadeUpstream: An intermediate replica a sub-replica may cascade from.
CausalBookmark
CommitWaiter
ControlPlanePayload: Opaque, already-encoded body of a control-plane entry.
DivergentTail: The divergent tail removed from the live timeline: the records in (common_point_lsn, to_lsn] that never reached quorum.
DownstreamSlot: A sub-replica slot held by an intermediate.
ElectionCoordinator: The quorum-gated election state machine.
ElectionRequest: A request to run an election on behalf of candidate.
FailoverCoordinator: The coordinated zero-RPO failover state machine.
FailoverNode: A node participating in a failover.
FailoverOutcome: The result of a completed handover.
FailoverRequest: A request to hand the primary role from old_primary to target.
FileLastVoteStore
FileTermStore
FlowController: Ticket-based write-admission flow controller.
LagConfig: Knobs for the lag/health computation. Kept as a small struct so the call sites (gRPC topology RPC, RedWire HelloAck builder) thread the same defaults without each one redeclaring constants.
LastVote: A node’s durable voting record: the highest term it has participated in and who, if anyone, it granted that term. Persisted so a restart cannot erase the fact that a vote was already cast (requirement 2).
LeaseStore: Wraps an AtomicRemoteBackend with lease primitives. The lease object is stored under a deterministic key derived from database_key; the store reads/writes that one key.
Member: A cluster member as seen by the supervisor’s membership view.
MemoryLastVoteStore: In-memory last-vote store for tests and witnesses that do not need cross-restart durability. (A witness should still persist in production; the file store is used there.)
MemoryTermStore: In-memory term store for tests and ephemeral nodes.
QuorumConfig: Quorum configuration stored alongside ReplicationConfig.
QuorumCoordinator: Tracks per-replica region bindings and pairs them with the primary’s ack map. PrimaryReplication owns the WAL buffer + ReplicaState list; this coordinator adds the region dimension and the wait-for- quorum logic without duplicating the ack table.
RebootstrapInProgress: A causal read was requested while the node is re-bootstrapping.
ReplicationConfig: Configuration for replication.
RoleAssignment: Post-handover roles of the two nodes, used to assert that the new primary advertises the new term and the old primary streams as a replica (issue #833 criterion 3).
RollbackCoordinator: The deposed-primary auto-rollback state machine.
RollbackEvent: The loud operator event payload describing a completed rollback, handed to RollbackTransport::emit_rollback_event. Mirrors crate::telemetry::operator_event::OperatorEvent::DeposedPrimaryRollback so the production transport can forward it verbatim while a test transport can capture it.
RollbackOutcome: The result of a completed rejoin.
RollbackPlan: The computed, side-effect-free rollback plan. Splitting this out lets the boundary invariant be asserted without driving any transport.
RollbackRequest: A request to auto-rollback a deposed primary to the common point and rejoin it as a replica.
StaleTermFenced: Why the term fence refused a message: the incoming term is behind the current term, so the sender is a deposed primary on a superseded timeline.
SwapDb: A dataset that stays readable across an atomic re-bootstrap swap.
TailRecord: A single record from the divergent tail that is about to be discarded.
TermFence: The stale-term fence. Wraps a durable TermStore and applies the term rule at the apply and handshake boundaries.
TopologyAdvertiser: Server-side advertiser. Zero-sized — all state is threaded through advertise()’s arguments so callers control the snapshot semantics.
TopologyAuthGate: Predicate over the caller’s auth context — answers “does this principal have cluster:topology:read?”.
VoteRequest: A request for a vote, sent by a candidate to a voter.
Voter: A voting member. Wraps the durable LastVoteStore and applies the vote rule. The voter is the seat of correctness: the watermark rule and the durable double-vote guard both live here.
WitnessSupervisor: A booted witness node: the control-plane supervisor with no data plane.
WriterLease

Enums§

Admission: Outcome of a write-admission attempt.
AwaitOutcome
BookmarkDecodeError
CascadeRefusal: Why a requested cascade source was refused and the node fell back to the primary. Surfaced (not swallowed) so a misconfiguration is observable rather than a silent performance cliff.
CommitPolicy
ControlPlaneEntry: A single entry in the control-plane log.
ControlPlaneEntryKind: The kind of a ControlPlaneEntry, without its payload.
ControlPlaneRole: This node’s role in the control-plane consensus layer.
ElectionOutcome: The result of an election attempt.
FailoverError: Why a coordinated failover could not complete without losing writes.
FailoverMode: How a failover should be executed.
FenceBoundary: The boundary at which a term-stamped message is being admitted. Only affects diagnostics — the term rule is identical at both.
FenceVerdict: The verdict of the term fence for one incoming term-stamped message.
LastVoteError
LeaseError
MemberKind: Whether a member holds data (and can therefore be promoted to primary) or is a vote-only witness (ADR 0030 — “a node that runs only the supervisor module”).
NodeRole: The replication role a node plays after a failover step.
ProposeRefusal: Why a ControlPlaneConsensus::propose was refused.
QuorumError: Errors raised by the quorum coordinator. The write itself succeeded on the primary WAL — these errors signal that replica acknowledgement did not reach quorum and the caller must decide whether to surface the failure or continue anyway.
RefusalReason: Why a voter refused a candidate.
ReplicaClass: How a node chooses its WAL upstream.
ReplicationRole: Role of this RedDB instance in a replication cluster.
RollbackError: Why an auto-rollback could not complete.
RuntimeProfile: Which planes a node boots.
TermStoreError: Error reading or persisting the durable current term.
UpstreamChoice: Where a node should open its WAL stream.
VoteDecision: The outcome of a voter considering a VoteRequest.
VotingState: Whether a member currently participates in voting.

Constants§

DEFAULT_REPLICATION_TERM
DEFAULT_REPLICA_TIMEOUT_MS: Default replica heartbeat timeout used when an operator hasn’t configured one explicitly. Matches the order of the poll_interval_ms default in ReplicationConfig (100 ms) multiplied by a generous fudge factor — five seconds without an ack flips a replica to healthy: false. Operators tune this via LagConfig.
DEFAULT_SLOT_IDLE_TIMEOUT_MS
DEFAULT_SLOT_RETENTION_MAX_LAG_LSN
TOPOLOGY_READ_CAPABILITY: Capability name from ADR 0008 §1.

Traits§

ControlPlaneConsensus: The Cluster Supervisor’s control-plane consensus seam.
ElectionTransport: Cluster operations the candidate drives, injected so the state machine stays pure and deterministically testable. Production backs these onto the membership view, the per-peer vote RPC, the durable term store, and the FAILOVER handover; tests back them onto a scripted fake.
FailoverTransport: Cluster mutations and the clock the coordinator drives, injected so the state machine stays pure and deterministically testable.
LastVoteStore: Durable store for a node’s last vote. The contract is narrow on purpose: load returns the persisted record (or the default term 0, voted_for None when nothing was ever written), and persist makes a record durable before the caller acknowledges a grant.
RollbackTransport: Side effects the rollback coordinator drives, injected so the state machine stays pure and deterministically testable.
TermStore: Durable store for a node’s current replication term. The default (when nothing was ever written) is DEFAULT_REPLICATION_TERM, matching the term records carry before any failover.

Functions§

plan_upstream: Decide where a node streams from, given its streaming class and an optionally-requested intermediate source.
quorum_threshold: Quorum threshold for a set of members: a strict majority of the voting members. Witnesses count; catching-up replicas do not.
randomized_election_timeout: A randomized election timeout in [base, base + jitter).

Type Aliases§

ControlPlaneLogIndex: Position of an entry in the control-plane log. Monotonic per term-history; an entry is durable once its index is at or below the committed index of a quorum.
MemberId: Stable identity of a cluster member, matching the election membership id (super::election::Member::id) and the replica/ack id namespace.

Module replication

Module replication Copy item path

§Architecture

§Usage

Modules§

Structs§

Enums§

Constants§

Traits§

Functions§

Type Aliases§

Module replication