fix cluster: hashRing stale after rebalance, GetShardMap empty, loops, and migration no-op

- GetShardMap: copy Shards and Nodes map contents before returning
- hashRing: rebuild from shardMap.Nodes after every rebalance/shard_map update
- DistributedVM.handleRebalanceRequest: add leader check and self-broadcast guard
- route_message: add hop-count (MaxRouteHops=10) to prevent infinite loops
- handleMigrationRequest: broadcast migration updates instead of setting local copy

cluster/distributed.go

@@ -150,6 +150,15 @@ func (dvm *DistributedVM) SendMessage(message RuntimeMessage) error {
 
 // routeMessageToNode sends a message to another node for delivery to the target actor
 func (dvm *DistributedVM) routeMessageToNode(actorID string, message RuntimeMessage) error {
+	hops := 0
+	if mp, ok := message.(*MessagePayload); ok {
+		hops = mp.Hops
+	}
+	if hops >= MaxRouteHops {
+		dvm.cluster.logger.Printf("Dropping message for actor %s: exceeded max hops (%d)", actorID, MaxRouteHops)
+		return fmt.Errorf("message exceeded max hops")
+	}
+
 	msg := ClusterMessage{
 		Type: "route_message",
 		From: dvm.nodeID,
@@ -157,6 +166,7 @@ func (dvm *DistributedVM) routeMessageToNode(actorID string, message RuntimeMess
 		Payload: MessagePayload{
 			TargetActorID: actorID,
 			Type:          message.GetType(),
+			Hops:          hops + 1,
 		},
 		Timestamp: time.Now(),
 	}
@@ -220,6 +230,11 @@ func (dvm *DistributedVM) handleClusterMessage(msg *nats.Msg) {
 			return
 		}
 
+		if message.Hops >= MaxRouteHops {
+			dvm.cluster.logger.Printf("Dropping message for actor %s: exceeded max hops (%d)", message.TargetActorID, MaxRouteHops)
+			return
+		}
+
 		targetActor := message.TargetActorID
 		if dvm.IsLocalActor(targetActor) {
 			dvm.localRuntime.SendMessage(&message)
@@ -236,6 +251,15 @@ func (dvm *DistributedVM) handleClusterMessage(msg *nats.Msg) {
 
 // handleRebalanceRequest processes shard rebalancing requests
 func (dvm *DistributedVM) handleRebalanceRequest(msg ClusterMessage) {
+	if msg.From == dvm.nodeID {
+		return
+	}
+
+	if !dvm.cluster.IsLeader() {
+		dvm.cluster.logger.Printf("Ignoring rebalance request: not the leader")
+		return
+	}
+
 	payloadBytes, err := json.Marshal(msg.Payload)
 	if err != nil {
 		dvm.cluster.logger.Printf("Failed to marshal rebalance payload: %v", err)
@@ -251,6 +275,10 @@ func (dvm *DistributedVM) handleRebalanceRequest(msg ClusterMessage) {
 	dvm.cluster.mutex.Lock()
 	if newShardMap.Version > dvm.cluster.shardMap.Version {
 		dvm.cluster.shardMap = &newShardMap
+		dvm.cluster.hashRing = NewConsistentHashRing()
+		for nodeID := range newShardMap.Nodes {
+			dvm.cluster.hashRing.AddNode(nodeID)
+		}
 		dvm.cluster.logger.Printf("Applied new shard map (version %d) from rebalance", newShardMap.Version)
 	} else {
 		dvm.cluster.logger.Printf("Ignoring stale shard map (got version %d, current %d)",

cluster/manager.go

@@ -156,6 +156,8 @@ func (cm *ClusterManager) handleClusterMessage(msg *nats.Msg) {
 		}
 	case "shard_map":
 		cm.handleShardMapUpdate(clusterMsg)
+	case "migration_update":
+		cm.handleMigrationUpdate(clusterMsg)
 	default:
 		cm.logger.Printf("Unknown cluster message type: %s", clusterMsg.Type)
 	}
@@ -249,6 +251,11 @@ func (cm *ClusterManager) handleRebalanceRequest(msg ClusterMessage) {
 	cm.shardMap = newShardMap
 	cm.mutex.Unlock()
 
+	cm.hashRing = NewConsistentHashRing()
+	for nodeID := range activeNodes {
+		cm.hashRing.AddNode(nodeID)
+	}
+
 	cm.broadcastShardMap(newShardMap)
 }
 
@@ -273,6 +280,31 @@ func (cm *ClusterManager) handleMigrationRequest(msg ClusterMessage) {
 	if migration.FromNode == cm.nodeID {
 		cm.logger.Printf("Initiating local actor state export for %s", migration.ActorID)
 		migration.Status = string(MigrationInProgress)
+		cm.broadcastMigrationUpdate(migration)
+	}
+
+	if migration.ToNode == cm.nodeID {
+		cm.logger.Printf("Actor %s assigned to this node, waiting for state import", migration.ActorID)
+	}
+}
+
+// broadcastMigrationUpdate propagates migration status updates to the cluster
+func (cm *ClusterManager) broadcastMigrationUpdate(migration ActorMigration) {
+	msg := ClusterMessage{
+		Type:    "migration_update",
+		From:    cm.nodeID,
+		To:      "broadcast",
+		Payload: migration,
+	}
+
+	data, err := json.Marshal(msg)
+	if err != nil {
+		cm.logger.Printf("Failed to marshal migration update: %v", err)
+		return
+	}
+
+	if err := cm.natsConn.Publish("aether.cluster.migration_update", data); err != nil {
+		cm.logger.Printf("Failed to publish migration update: %v", err)
 	}
 }
 
@@ -309,6 +341,11 @@ func (cm *ClusterManager) triggerShardRebalancing(reason string) {
 	cm.shardMap = newShardMap
 	cm.mutex.Unlock()
 
+	cm.hashRing = NewConsistentHashRing()
+	for nodeID := range activeNodes {
+		cm.hashRing.AddNode(nodeID)
+	}
+
 	cm.broadcastShardMap(newShardMap)
 }
 
@@ -396,6 +433,10 @@ func (cm *ClusterManager) handleShardMapUpdate(msg ClusterMessage) {
 	cm.mutex.Lock()
 	if newShardMap.Version > cm.shardMap.Version {
 		cm.shardMap = &newShardMap
+		cm.hashRing = NewConsistentHashRing()
+		for nodeID := range newShardMap.Nodes {
+			cm.hashRing.AddNode(nodeID)
+		}
 		cm.logger.Printf("Applied new shard map (version %d)", newShardMap.Version)
 	} else {
 		cm.logger.Printf("Ignoring stale shard map (got version %d, current %d)",
@@ -410,12 +451,22 @@ func (cm *ClusterManager) GetShardMap() *ShardMap {
 	defer cm.mutex.RUnlock()
 
 	// Return a copy to prevent external mutation
-	return &ShardMap{
+	copy := &ShardMap{
 		Version:    cm.shardMap.Version,
 		Shards:     make(map[int][]string),
 		Nodes:      make(map[string]NodeInfo),
 		UpdateTime: cm.shardMap.UpdateTime,
 	}
+
+	for shardID, nodes := range cm.shardMap.Shards {
+		copy.Shards[shardID] = append([]string(nil), nodes...)
+	}
+
+	for nodeID, nodeInfo := range cm.shardMap.Nodes {
+		copy.Nodes[nodeID] = nodeInfo
+	}
+
+	return copy
 }
 
 // broadcastShardMap propagates a new shard map to all cluster nodes via NATS
@@ -440,3 +491,20 @@ func (cm *ClusterManager) broadcastShardMap(newShardMap *ShardMap) {
 
 	cm.logger.Printf("Broadcast new shard map (version %d) to cluster", newShardMap.Version)
 }
+
+// handleMigrationUpdate processes migration status update messages from other nodes
+func (cm *ClusterManager) handleMigrationUpdate(msg ClusterMessage) {
+	var migration ActorMigration
+	payloadBytes, err := json.Marshal(msg.Payload)
+	if err != nil {
+		cm.logger.Printf("Failed to marshal migration update payload: %v", err)
+		return
+	}
+	if err := json.Unmarshal(payloadBytes, &migration); err != nil {
+		cm.logger.Printf("Failed to unmarshal migration update: %v", err)
+		return
+	}
+
+	cm.logger.Printf("Migration update for actor %s: status=%s (from %s)",
+		migration.ActorID, migration.Status, msg.From)
+}

cluster/types.go

@@ -191,11 +191,15 @@ func (m *ModelPayload) GetID() string { return m.ID }
 // GetName implements RuntimeModel
 func (m *ModelPayload) GetName() string { return m.Name }
 
+// MaxRouteHops is the maximum number of hops a routed message can take before being dropped
+const MaxRouteHops = 10
+
 // MessagePayload is a concrete type for JSON-unmarshaling RuntimeMessage payloads.
 // Use this when receiving message data over the network.
 type MessagePayload struct {
 	TargetActorID string `json:"targetActorId"`
 	Type          string `json:"type"`
+	Hops          int    `json:"hops,omitempty"`
 }
 
 // GetTargetActorID implements RuntimeMessage