@feltdb/core 0.4.11 → 0.4.13

package/dist/create/server-source/crates/feltdb/src/replay_fuzzing.rs

@@ -0,0 +1,401 @@
+/// Replay Fuzzing Framework
+///
+/// Tests that the same operation log produces identical state when replayed via:
+/// - Normal replay (in original order)
+/// - Batched replay (grouped operations)
+/// - After snapshot (recover from snapshot + remaining ops)
+/// - After simulated crash (recover + replay remaining)
+/// - After index deletion (rebuild index from scratch)
+///
+/// All paths must converge to identical state_hash.
+
+use crate::operation_log::{MemoryOperationLog, OperationLogEntry};
+use crate::state_hash::{CanonicalState, StateHash};
+use crate::state_transition_store::MemoryStateTransitionStore;
+use crate::transactions::*;
+use std::collections::HashMap;
+
+/// Replay scenario
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub enum ReplayScenario {
+    /// Replay all operations in original order
+    Normal,
+    /// Replay in batches of N operations
+    Batched { batch_size: usize },
+    /// Simulate recovery from snapshot at index N, replay rest
+    AfterSnapshot { snapshot_at: u64 },
+    /// Simulate crash after index N, recover and replay
+    AfterCrash { crash_at: u64 },
+    /// Simulate index deletion, rebuild from log
+    IndexRebuilt,
+}
+
+/// Replay path and result
+#[derive(Clone, Debug)]
+pub struct ReplayPath {
+    pub scenario: ReplayScenario,
+    pub final_state_hash: StateHash,
+    pub operations_replayed: u64,
+    pub transitions_created: u64,
+}
+
+/// Replay fuzzer
+pub struct ReplayFuzzer {
+    operation_log: MemoryOperationLog,
+    transition_store: MemoryStateTransitionStore,
+}
+
+impl ReplayFuzzer {
+    pub fn new(operation_log: MemoryOperationLog, transition_store: MemoryStateTransitionStore) -> Self {
+        Self {
+            operation_log,
+            transition_store,
+        }
+    }
+
+    /// Replay operations normally (original order)
+    pub fn replay_normal(&self) -> ReplayPath {
+        let all_ops = self.operation_log.iter();
+        let mut state = CanonicalState::new();
+
+        for entry in all_ops {
+            self.apply_operation_to_state(&mut state, &entry.operation);
+        }
+
+        ReplayPath {
+            scenario: ReplayScenario::Normal,
+            final_state_hash: state.hash(),
+            operations_replayed: self.operation_log.size(),
+            transitions_created: 1,
+        }
+    }
+
+    /// Replay operations in batches
+    pub fn replay_batched(&self, batch_size: usize) -> ReplayPath {
+        let mut state = CanonicalState::new();
+        let log_size = self.operation_log.size();
+        let mut total_batches = 0;
+
+        for batch_start in (0..log_size).step_by(batch_size) {
+            let batch_end = std::cmp::min(batch_start + batch_size as u64, log_size);
+            let batch = self.operation_log.get_range(batch_start, batch_end);
+
+            for entry in batch {
+                self.apply_operation_to_state(&mut state, &entry.operation);
+            }
+            total_batches += 1;
+        }
+
+        ReplayPath {
+            scenario: ReplayScenario::Batched { batch_size },
+            final_state_hash: state.hash(),
+            operations_replayed: self.operation_log.size(),
+            transitions_created: total_batches as u64,
+        }
+    }
+
+    /// Replay from snapshot (skip first N operations, replay rest)
+    pub fn replay_after_snapshot(&self, snapshot_at: u64) -> ReplayPath {
+        let log_size = self.operation_log.size();
+        if snapshot_at >= log_size {
+            return ReplayPath {
+                scenario: ReplayScenario::AfterSnapshot {
+                    snapshot_at,
+                },
+                final_state_hash: StateHash::from_hex("snapshot_beyond_log".to_string()),
+                operations_replayed: 0,
+                transitions_created: 0,
+            };
+        }
+
+        let mut state = CanonicalState::new();
+
+        // Simulate snapshot state (in real system, would be loaded from snapshot)
+        // For this test, assume snapshot captures state after snapshot_at operations
+        let pre_snapshot_ops = self.operation_log.get_range(0, snapshot_at);
+        for entry in pre_snapshot_ops {
+            self.apply_operation_to_state(&mut state, &entry.operation);
+        }
+
+        // Replay remaining operations
+        let remaining_ops = self.operation_log.get_range(snapshot_at, log_size);
+        for entry in remaining_ops {
+            self.apply_operation_to_state(&mut state, &entry.operation);
+        }
+
+        ReplayPath {
+            scenario: ReplayScenario::AfterSnapshot { snapshot_at },
+            final_state_hash: state.hash(),
+            operations_replayed: log_size,
+            transitions_created: 1,
+        }
+    }
+
+    /// Replay after simulated crash (skip operations after crash point)
+    pub fn replay_after_crash(&self, crash_at: u64) -> ReplayPath {
+        let log_size = self.operation_log.size();
+        let replay_limit = std::cmp::min(crash_at, log_size);
+
+        let mut state = CanonicalState::new();
+        let ops = self.operation_log.get_range(0, replay_limit);
+
+        for entry in ops {
+            self.apply_operation_to_state(&mut state, &entry.operation);
+        }
+
+        ReplayPath {
+            scenario: ReplayScenario::AfterCrash { crash_at },
+            final_state_hash: state.hash(),
+            operations_replayed: replay_limit,
+            transitions_created: 1,
+        }
+    }
+
+    /// Replay with index rebuilt from log
+    pub fn replay_index_rebuilt(&self) -> ReplayPath {
+        // Clear transition store (simulate index deletion)
+        let all_ops = self.operation_log.iter();
+        let mut state = CanonicalState::new();
+
+        for entry in all_ops {
+            self.apply_operation_to_state(&mut state, &entry.operation);
+        }
+
+        ReplayPath {
+            scenario: ReplayScenario::IndexRebuilt,
+            final_state_hash: state.hash(),
+            operations_replayed: self.operation_log.size(),
+            transitions_created: 1,
+        }
+    }
+
+    /// Apply an operation's effects to state
+    fn apply_operation_to_state(&self, state: &mut CanonicalState, operation: &Operation) {
+        if operation.command.op_type == "UpdateField" {
+            let collection = operation.command.collection.clone();
+            let record_id = operation.command.record_id.clone();
+            let data = serde_json::json!(operation.command.fields.clone());
+            state.set_record(collection, record_id, data);
+        }
+    }
+
+    /// Execute full replay scenarios (all operations replayed)
+    pub fn execute_full_scenarios(&self) -> Vec<ReplayPath> {
+        vec![
+            self.replay_normal(),
+            self.replay_batched(2),
+            self.replay_batched(5),
+            self.replay_after_snapshot(self.operation_log.size() / 2),
+            self.replay_index_rebuilt(),
+        ]
+    }
+
+    /// Execute all scenarios (including partial)
+    pub fn execute_all_scenarios(&self) -> Vec<ReplayPath> {
+        let mut all = self.execute_full_scenarios();
+        all.push(self.replay_after_crash(self.operation_log.size().saturating_sub(1)));
+        all
+    }
+
+    /// Verify all FULL replay paths converge to same state
+    pub fn verify_convergence(&self) -> Result<StateHash, String> {
+        let paths = self.execute_full_scenarios();
+
+        if paths.is_empty() {
+            return Err("No replay paths generated".to_string());
+        }
+
+        let canonical_hash = &paths[0].final_state_hash;
+
+        for path in &paths[1..] {
+            if &path.final_state_hash != canonical_hash {
+                return Err(format!(
+                    "Convergence failed: {:?} produced {}, expected {}",
+                    path.scenario, path.final_state_hash, canonical_hash
+                ));
+            }
+        }
+
+        Ok(canonical_hash.clone())
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::convergence::VectorClock;
+
+    fn create_test_operation(seq: u64, collection: &str, record_id: &str, value: i32) -> Operation {
+        let mut vc = VectorClock::new();
+        vc.increment("replay_test");
+        let version = StateVersion::new(vc, "test_hash".to_string());
+
+        Operation::new(
+            OperationId::new("replay_test".to_string(), seq),
+            version,
+            format!("Op{}", seq),
+            OperationCommand {
+                op_type: "UpdateField".to_string(),
+                collection: collection.to_string(),
+                record_id: record_id.to_string(),
+                fields: {
+                    let mut m = HashMap::new();
+                    m.insert("value".to_string(), serde_json::json!(value));
+                    m
+                },
+            },
+            "test".to_string(),
+        )
+    }
+
+    #[test]
+    fn test_fuzzer_creation() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+        let _fuzzer = ReplayFuzzer::new(log, store);
+    }
+
+    #[test]
+    fn test_replay_normal_empty_log() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+        let fuzzer = ReplayFuzzer::new(log, store);
+
+        let result = fuzzer.replay_normal();
+        assert_eq!(result.operations_replayed, 0);
+    }
+
+    #[test]
+    fn test_replay_normal_single_operation() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        let op = create_test_operation(1, "records", "rec1", 100);
+        log.append(op).unwrap();
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.replay_normal();
+
+        assert_eq!(result.operations_replayed, 1);
+        assert_eq!(result.scenario, ReplayScenario::Normal);
+    }
+
+    #[test]
+    fn test_replay_batched() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=5 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.replay_batched(2);
+
+        assert_eq!(result.operations_replayed, 5);
+        assert_eq!(result.scenario, ReplayScenario::Batched { batch_size: 2 });
+    }
+
+    #[test]
+    fn test_replay_after_snapshot() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=5 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.replay_after_snapshot(3);
+
+        assert_eq!(result.operations_replayed, 5);
+    }
+
+    #[test]
+    fn test_replay_after_crash() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=5 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.replay_after_crash(3);
+
+        assert_eq!(result.operations_replayed, 3);
+    }
+
+    #[test]
+    fn test_replay_index_rebuilt() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=3 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.replay_index_rebuilt();
+
+        assert_eq!(result.operations_replayed, 3);
+        assert_eq!(result.scenario, ReplayScenario::IndexRebuilt);
+    }
+
+    #[test]
+    fn test_all_scenarios_same_hash() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=5 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+
+        let normal = fuzzer.replay_normal();
+        let batched = fuzzer.replay_batched(2);
+        let snapshot = fuzzer.replay_after_snapshot(2);
+
+        assert_eq!(normal.final_state_hash, batched.final_state_hash);
+        assert_eq!(normal.final_state_hash, snapshot.final_state_hash);
+    }
+
+    #[test]
+    fn test_verify_convergence() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=5 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let result = fuzzer.verify_convergence();
+
+        assert!(result.is_ok());
+    }
+
+    #[test]
+    fn test_execute_all_scenarios() {
+        let log = MemoryOperationLog::new();
+        let store = MemoryStateTransitionStore::new();
+
+        for i in 1..=3 {
+            let op = create_test_operation(i, "records", &format!("rec{}", i), i as i32 * 10);
+            log.append(op).unwrap();
+        }
+
+        let fuzzer = ReplayFuzzer::new(log, store);
+        let paths = fuzzer.execute_all_scenarios();
+
+        assert!(paths.len() > 0);
+        assert!(paths.iter().all(|p| p.operations_replayed > 0));
+    }
+}

package/dist/create/server-source/crates/feltdb/src/replication_manager.rs

@@ -0,0 +1,160 @@
+/// Replication Manager
+///
+/// Coordinates message routing and delivery between nodes in a distributed cluster.
+/// Handles:
+/// - Tracking pending operations and their delivery status
+/// - Routing messages from leader to followers
+/// - Acknowledgment tracking
+/// - Retry logic for lost messages
+
+use crate::distributed_transactions::ReplicationMessage;
+use std::collections::{HashMap, VecDeque};
+use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::Arc;
+
+/// Message delivery status
+#[derive(Clone, Debug, PartialEq)]
+pub enum DeliveryStatus {
+    Pending,
+    Delivered,
+    Acknowledged,
+    Failed,
+}
+
+/// Tracked message for replication
+#[derive(Clone)]
+pub struct TrackedMessage {
+    pub message: ReplicationMessage,
+    pub status: DeliveryStatus,
+    pub send_count: u64,
+    pub timestamp_ms: u64,
+}
+
+impl TrackedMessage {
+    pub fn new(message: ReplicationMessage) -> Self {
+        let timestamp_ms = std::time::SystemTime::now()
+            .duration_since(std::time::UNIX_EPOCH)
+            .unwrap()
+            .as_millis() as u64;
+
+        Self {
+            message,
+            status: DeliveryStatus::Pending,
+            send_count: 0,
+            timestamp_ms,
+        }
+    }
+}
+
+/// Tracks pending messages to be replicated to a specific peer
+pub struct PeerReplicationQueue {
+    peer_id: String,
+    messages: VecDeque<TrackedMessage>,
+    next_seq: Arc<AtomicU64>,
+}
+
+impl PeerReplicationQueue {
+    pub fn new(peer_id: String) -> Self {
+        Self {
+            peer_id,
+            messages: VecDeque::new(),
+            next_seq: Arc::new(AtomicU64::new(1)),
+        }
+    }
+
+    pub fn enqueue(&mut self, message: ReplicationMessage) {
+        self.messages.push_back(TrackedMessage::new(message));
+    }
+
+    pub fn pending_count(&self) -> usize {
+        self.messages.iter().filter(|m| m.status == DeliveryStatus::Pending).count()
+    }
+
+    pub fn next_pending(&mut self) -> Option<TrackedMessage> {
+        for msg in &mut self.messages {
+            if msg.status == DeliveryStatus::Pending {
+                msg.send_count += 1;
+                return Some(msg.clone());
+            }
+        }
+        None
+    }
+
+    pub fn mark_delivered(&mut self, seq: u64) {
+        if seq > 0 && seq as usize <= self.messages.len() {
+            if let Some(msg) = self.messages.get_mut((seq - 1) as usize) {
+                msg.status = DeliveryStatus::Delivered;
+            }
+        }
+    }
+}
+
+/// Replication manager for a cluster
+pub struct ReplicationManager {
+    peer_queues: HashMap<String, PeerReplicationQueue>,
+}
+
+impl ReplicationManager {
+    pub fn new() -> Self {
+        Self {
+            peer_queues: HashMap::new(),
+        }
+    }
+
+    /// Register a peer to receive replication messages
+    pub fn register_peer(&mut self, peer_id: String) {
+        self.peer_queues.insert(peer_id.clone(), PeerReplicationQueue::new(peer_id));
+    }
+
+    /// Enqueue a message for replication to all peers
+    pub fn replicate_to_all(&mut self, message: ReplicationMessage) {
+        for queue in self.peer_queues.values_mut() {
+            queue.enqueue(message.clone());
+        }
+    }
+
+    /// Enqueue a message for replication to a specific peer
+    pub fn replicate_to_peer(&mut self, peer_id: &str, message: ReplicationMessage) {
+        if let Some(queue) = self.peer_queues.get_mut(peer_id) {
+            queue.enqueue(message);
+        }
+    }
+
+    /// Get pending messages for a peer
+    pub fn get_pending_for_peer(&mut self, peer_id: &str) -> Vec<TrackedMessage> {
+        let mut result = Vec::new();
+        if let Some(queue) = self.peer_queues.get_mut(peer_id) {
+            while let Some(msg) = queue.next_pending() {
+                result.push(msg);
+            }
+        }
+        result
+    }
+
+    /// Mark a message as delivered to a peer
+    pub fn mark_peer_delivered(&mut self, peer_id: &str, seq: u64) {
+        if let Some(queue) = self.peer_queues.get_mut(peer_id) {
+            queue.mark_delivered(seq);
+        }
+    }
+
+    /// Get replication statistics
+    pub fn get_pending_count(&self) -> usize {
+        self.peer_queues.values().map(|q| q.pending_count()).sum()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn replication_manager_basic() {
+        let mut mgr = ReplicationManager::new();
+        mgr.register_peer("peer1".to_string());
+
+        // Verify peer registered
+        assert_eq!(mgr.peer_queues.len(), 1);
+        assert_eq!(mgr.get_pending_count(), 0);
+    }
+}

package/dist/create/server-source/crates/feltdb/src/replication_protocol.rs

@@ -0,0 +1,132 @@
+/// High-level replication protocol abstraction
+///
+/// Decouples ReplicationMessage handling from transport details.
+/// Works with typed messages (ReplicationMessage) rather than raw bytes.
+
+use crate::distributed_transactions::ReplicationMessage;
+use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::Arc;
+
+/// Protocol-level transport abstraction that works with ReplicationMessage
+#[async_trait::async_trait]
+pub trait ProtocolTransport: Send + Sync {
+    /// Send a replication message to a peer
+    async fn send(&self, message: ReplicationMessage) -> Result<(), ProtocolError>;
+
+    /// Receive replication messages (non-blocking, returns empty if none available)
+    async fn receive(&self) -> Result<Vec<ReplicationMessage>, ProtocolError>;
+
+    /// Connect to a peer (for real network transports)
+    async fn connect(&mut self, peer_addr: &str) -> Result<(), ProtocolError>;
+
+    /// Disconnect from a peer (for real network transports)
+    async fn disconnect(&mut self) -> Result<(), ProtocolError>;
+
+    /// Check if connected to a peer
+    fn is_connected(&self) -> bool;
+}
+
+/// Protocol errors
+#[derive(Clone, Debug)]
+pub enum ProtocolError {
+    SerializationError(String),
+    TransportError(String),
+    ConnectionError(String),
+    Timeout,
+}
+
+impl std::fmt::Display for ProtocolError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::SerializationError(s) => write!(f, "serialization error: {}", s),
+            Self::TransportError(s) => write!(f, "transport error: {}", s),
+            Self::ConnectionError(s) => write!(f, "connection error: {}", s),
+            Self::Timeout => write!(f, "timeout"),
+        }
+    }
+}
+
+impl std::error::Error for ProtocolError {}
+
+/// Metrics for monitoring replication protocol
+#[derive(Clone, Debug, Default)]
+pub struct ProtocolMetrics {
+    messages_sent: Arc<AtomicU64>,
+    messages_received: Arc<AtomicU64>,
+    bytes_sent: Arc<AtomicU64>,
+    bytes_received: Arc<AtomicU64>,
+    errors: Arc<AtomicU64>,
+    connection_state_changes: Arc<AtomicU64>,
+}
+
+impl ProtocolMetrics {
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    pub fn record_send(&self, bytes: usize) {
+        self.messages_sent.fetch_add(1, Ordering::Relaxed);
+        self.bytes_sent.fetch_add(bytes as u64, Ordering::Relaxed);
+    }
+
+    pub fn record_receive(&self, bytes: usize) {
+        self.messages_received.fetch_add(1, Ordering::Relaxed);
+        self.bytes_received.fetch_add(bytes as u64, Ordering::Relaxed);
+    }
+
+    pub fn record_error(&self) {
+        self.errors.fetch_add(1, Ordering::Relaxed);
+    }
+
+    pub fn record_connection_state_change(&self) {
+        self.connection_state_changes.fetch_add(1, Ordering::Relaxed);
+    }
+
+    pub fn get_messages_sent(&self) -> u64 {
+        self.messages_sent.load(Ordering::Relaxed)
+    }
+
+    pub fn get_messages_received(&self) -> u64 {
+        self.messages_received.load(Ordering::Relaxed)
+    }
+
+    pub fn get_bytes_sent(&self) -> u64 {
+        self.bytes_sent.load(Ordering::Relaxed)
+    }
+
+    pub fn get_bytes_received(&self) -> u64 {
+        self.bytes_received.load(Ordering::Relaxed)
+    }
+
+    pub fn get_errors(&self) -> u64 {
+        self.errors.load(Ordering::Relaxed)
+    }
+
+    pub fn get_connection_state_changes(&self) -> u64 {
+        self.connection_state_changes.load(Ordering::Relaxed)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn metrics_track_messages() {
+        let metrics = ProtocolMetrics::new();
+        metrics.record_send(1024);
+        metrics.record_send(2048);
+
+        assert_eq!(metrics.get_messages_sent(), 2);
+        assert_eq!(metrics.get_bytes_sent(), 3072);
+    }
+
+    #[test]
+    fn metrics_track_errors() {
+        let metrics = ProtocolMetrics::new();
+        metrics.record_error();
+        metrics.record_error();
+
+        assert_eq!(metrics.get_errors(), 2);
+    }
+}