Added support for multi-node config settings

2026-10-15 17:11:30 +00:00 · 2025-11-16 21:20:08 +08:00
parent bacd1f3ba2
commit ff22a08948
11 changed files with 162 additions and 181 deletions
@@ -53,8 +53,10 @@ Confirm ./bin/hive exists and executable.

 When running Hive for the first time, run the init function to create the default directories for data, config files and layout required.

-```bash
-./bin/hive admin init
+Example single node installation to get started:
+
+```
+/bin/hive admin init --region ap-southeast-2 --az ap-southeast-2a --node node1 --nodes 1
 ```

 Next, set the AWS profile to use `hive` which points to the local environment.
@@ -93,7 +95,7 @@ aws ec2 create-key-pair \
 | jq -r '.KeyMaterial | rtrimstr("\n")' > ~/.ssh/hive-key
 ```

-Update permissions to SSH will accept reading the file.
+Update permissions for the key for ssh to accept reading the file.

 ```bash
 chmod 600 ~/.ssh/hive-key
@@ -105,7 +107,7 @@ Next, generate a public key from the specified private key pair.
 ssh-keygen -y -f ~/.ssh/hive-key > ~/.ssh/hive-key.pub
 ```

-Validate the new key is available
+### Validate the new key is available

 ```bash
 aws ec2 describe-key-pairs
@@ -164,13 +166,13 @@ Extracted image to: /home/ben/hive/images/debian/12/arm64/disk.raw
 ✅ Image import complete. Image-ID (AMI): ami-e29fcc65734aec9ea
 ```

-Next, verify available disk images to confirm the import was successful
+Next, verify available disk images to confirm the import was successful, replace `ami-XXX` with the value from the command output.

 ```bash
-aws ec2 describe-images --image-ids ami-e29fcc65734aec9ea
+aws ec2 describe-images --image-ids ami-XXX
 ```

-### Manual import
+### Manual AMI import

 Using this method you can import any OS disk image. For example, download the Debian 12 image from the repository [https://cloud.debian.org/images/cloud/bookworm/latest/](https://cloud.debian.org/images/cloud/bookworm/latest/)

@@ -26,22 +26,22 @@ Design layout for multi-node configuration.

 ```
 # node1
-hive admin init --region ap-southeast-2 --availability-zone ap-southeast-2a --nodes 3 --bind :8443 --data-dir ~/hive/
+hive admin init --region ap-southeast-2 --az ap-southeast-2a --node node1 --nodes 1 --hive-dir ~/hive/

 # node2
-hive admin join --region ap-southeast-2 --availability-zone ap-southeast-2a --host node1.local:8443 --data-dir ~/hive/
+hive admin join --region ap-southeast-2 --az ap-southeast-2a --node node2 --host node1.local:8443 --data-dir ~/hive/

 # node3, optionally toggle EBS/EC2/NATs support only
-hive admin join --region ap-southeast-2 --availability-zone ap-southeast-2a --host node1.local:8443 --data-dir ~/hive/ --cap ebs,ec2,nats
+hive admin join --region ap-southeast-2 --az ap-southeast-2a --node node3 --host node1.local:8443 --data-dir ~/hive/ --cap ebs,ec2,nats
 ```

-If --host is missing, the join command tries multicast broadcast to find parent node.
+If --host is missing, the join command tries multicast broadcast to find parent (leader) node.

 Set region with `--region ap-southeast-2` which will create a new Hive cluster for the specified region.

-Set the availability-zone, for production it is recommended to have at least 2 availability-zones defined per region deployment of Hive.
+For production, it is recommended to run Hive on at least three physical nodes. In this minimum setup, Hive S3 and EBS volumes use 2 data shards and 1 parity shard RS(2,1) to distribute each chunk across the cluster. This configuration tolerates a full node failure while adding only 1.5x storage overhead.

-Overview:
+### Overview:

 * init creates a cluster-id, node-id, and a short join token, starts a tiny control server on :8443, and writes DNS hint files if available.

@@ -140,6 +140,11 @@ func init() {
 	// Flags for admin init
 	adminInitCmd.Flags().Bool("force", false, "Force re-initialization (overwrites existing config)")
 	adminInitCmd.Flags().String("region", "ap-southeast-2", "Mulga region to create")
+	adminInitCmd.Flags().String("az", "ap-southeast-2a", "Mulga AZ to create")
+	adminInitCmd.Flags().String("node", "node1", "Node name, increment for additional nodes (default, node1)")
+	adminInitCmd.Flags().Int("nodes", 3, "Number of nodes to expect for cluster")
+	adminInitCmd.Flags().String("host", "", "Leader node to join (if not specified, tries multicast discovery)")
+	adminInitCmd.Flags().Int("port", 4432, "Port to bind cluster services on")

 	imagesImportCmd.Flags().String("tmp-dir", os.TempDir(), "Temporary directory for image import processing")

@@ -310,7 +315,7 @@ func runimagesImportCmd(cmd *cobra.Command, args []string) {
 	// Calculate the size

 	manifest.AMIMetadata.Name = fmt.Sprintf("ami-%s-%s-%s", image.Distro, image.Version, image.Arch)
-	volumeId := viperblock.GenerateVolumeID("ami", manifest.AMIMetadata.Name, appConfig.Predastore.Bucket, time.Now().Unix())
+	volumeId := viperblock.GenerateVolumeID("ami", manifest.AMIMetadata.Name, "", time.Now().Unix()) // TODO: Replace with bucket
 	manifest.AMIMetadata.ImageID = volumeId

 	manifest.AMIMetadata.Description = fmt.Sprintf("%s cloud image prepared for Hive", manifest.AMIMetadata.Name)
@@ -361,11 +366,11 @@ func runimagesImportCmd(cmd *cobra.Command, args []string) {
 	s3Config := s3.S3Config{
 		VolumeName: volumeId,
 		VolumeSize: uint64(imageStat.Size()),
-		Bucket:     appConfig.Predastore.Bucket,
-		Region:     appConfig.Predastore.Region,
-		AccessKey:  appConfig.AccessKey,
-		SecretKey:  appConfig.SecretKey,
-		Host:       appConfig.Predastore.Host,
+		Bucket:     appConfig.Nodes[appConfig.Node].Predastore.Bucket,
+		Region:     appConfig.Nodes[appConfig.Node].Predastore.Region,
+		AccessKey:  appConfig.Nodes[appConfig.Node].AccessKey,
+		SecretKey:  appConfig.Nodes[appConfig.Node].SecretKey,
+		Host:       appConfig.Nodes[appConfig.Node].Predastore.Host,
 	}

 	vbConfig := viperblock.VB{
@@ -435,6 +440,11 @@ func runAdminInit(cmd *cobra.Command, args []string) {
 	force, _ := cmd.Flags().GetBool("force")
 	configDir, _ := cmd.Flags().GetString("config-dir")
 	region, _ := cmd.Flags().GetString("region")
+	az, _ := cmd.Flags().GetString("az")
+	node, _ := cmd.Flags().GetString("node")
+	port, _ := cmd.Flags().GetInt("port")
+
+	fmt.Println("Config", az, node, port)

 	// Default config directory
 	if configDir == "" {
@@ -529,8 +539,10 @@ func runAdminInit(cmd *cobra.Command, args []string) {
 		{filepath.Join(natsDir, "nats.conf"), natsConfTemplate, "nats/nats.conf"},
 	}

+	portStr := fmt.Sprintf("%d", port)
+
 	for _, cfg := range configs {
-		if err := generateConfigFile(cfg.path, cfg.template, accessKey, secretKey, accountID, region, natsToken, hiveRoot); err != nil {
+		if err := generateConfigFile(cfg.path, cfg.template, accessKey, secretKey, accountID, region, natsToken, hiveRoot, az, node, portStr); err != nil {
 			fmt.Fprintf(os.Stderr, "Error creating %s: %v\n", cfg.name, err)
 			os.Exit(1)
 		}
@@ -693,7 +705,7 @@ func generateSelfSignedCert(certPath, keyPath string) error {
 }

 // generateConfigFile creates a configuration file from a template
-func generateConfigFile(path, templateContent, accessKey, secretKey, accountID, region, natsToken, dataDir string) error {
+func generateConfigFile(path, templateContent, accessKey, secretKey, accountID, region, natsToken, dataDir, Az, Node, Port string) error {
 	// Parse the embedded template
 	tmpl, err := template.New("config").Parse(templateContent)
 	if err != nil {
@@ -708,6 +720,11 @@ func generateConfigFile(path, templateContent, accessKey, secretKey, accountID,
 		Region    string
 		NatsToken string
 		DataDir   string
+
+		// Add more fields as needed
+		Node string
+		Az   string
+		Port string
 	}{
 		AccessKey: accessKey,
 		SecretKey: secretKey,
@@ -715,6 +732,10 @@ func generateConfigFile(path, templateContent, accessKey, secretKey, accountID,
 		Region:    region,
 		NatsToken: natsToken,
 		DataDir:   dataDir,
+
+		Node: Node,
+		Az:   Az,
+		Port: Port,
 	}

 	// Create file with secure permissions
@@ -26,7 +26,7 @@ import (

 var (
 	cfgFile   string
-	appConfig *config.Config
+	appConfig *config.ClusterConfig
 )

 // rootCmd represents the base command when called without any subcommands
@@ -167,7 +167,8 @@ var viperblockStartCmd = &cobra.Command{
 		fmt.Println("Loading config from:", cfgFile)

 		// TODO: Support ENV vars, CLI, otherwise revert to config.LoadConfig()
-		appConfig, err := config.LoadConfig(cfgFile)
+		clusterConfig, err := config.LoadConfig(cfgFile)
+		nodeConfig := clusterConfig.Nodes[clusterConfig.Node]

 		if err != nil {
 			fmt.Println("Error loading config file:", err)
@@ -178,48 +179,51 @@ var viperblockStartCmd = &cobra.Command{

 		if natsHost != "" {
 			fmt.Println("Overwriting natsHost:", natsHost)
-			appConfig.NATS.Host = natsHost
+			nodeConfig.NATS.Host = natsHost
 		}

 		s3Host := viper.GetString("s3-host")

 		if s3Host != "" {
 			fmt.Println("Overwriting s3host:", s3Host)
-			appConfig.Predastore.Host = s3Host
+			nodeConfig.Predastore.Host = s3Host
 		}

 		s3Bucket := viper.GetString("s3-bucket")

 		if s3Bucket != "" {
 			fmt.Println("Overwriting s3bucket:", s3Bucket)
-			appConfig.Predastore.Bucket = s3Bucket
+			nodeConfig.Predastore.Bucket = s3Bucket
 		}

 		s3Region := viper.GetString("s3-region")

 		if s3Region != "" {
 			fmt.Println("Overwriting s3Region:", s3Region)
-			appConfig.Predastore.Region = s3Region
+			nodeConfig.Predastore.Region = s3Region
 		}

 		accessKey := viper.GetString("access-key")
 		if accessKey != "" {
 			fmt.Println("Overwriting access-key:", accessKey)
-			appConfig.Predastore.AccessKey = accessKey
+			nodeConfig.Predastore.AccessKey = accessKey
 		}

 		secretKey := viper.GetString("secret-key")
 		if secretKey != "" {
 			fmt.Println("Overwriting secret-key:", secretKey)
-			appConfig.Predastore.SecretKey = secretKey
+			nodeConfig.Predastore.SecretKey = secretKey
 		}

 		baseDir := viper.GetString("base-dir")
 		if baseDir != "" {
 			fmt.Println("Overwriting base-dir:", baseDir)
-			appConfig.Predastore.BaseDir = baseDir
+			nodeConfig.Predastore.BaseDir = baseDir
 		}

+		// Apply changes back to cluster config
+		clusterConfig.Nodes[clusterConfig.Node] = nodeConfig
+
 		pluginPath := viper.GetString("plugin-path")

 		if pluginPath == "" {
@@ -236,14 +240,14 @@ var viperblockStartCmd = &cobra.Command{
 		}

 		service, err := service.New("viperblock", &viperblockd.Config{
-			NatsHost:   appConfig.NATS.Host,
+			NatsHost:   nodeConfig.NATS.Host,
 			PluginPath: pluginPath,
-			S3Host:     appConfig.Predastore.Host,
-			Bucket:     appConfig.Predastore.Bucket,
-			Region:     appConfig.Predastore.Region,
-			AccessKey:  appConfig.Predastore.AccessKey,
-			SecretKey:  appConfig.Predastore.SecretKey,
-			BaseDir:    appConfig.Predastore.BaseDir,
+			S3Host:     nodeConfig.Predastore.Host,
+			Bucket:     nodeConfig.Predastore.Bucket,
+			Region:     nodeConfig.Predastore.Region,
+			AccessKey:  nodeConfig.Predastore.AccessKey,
+			SecretKey:  nodeConfig.Predastore.SecretKey,
+			BaseDir:    nodeConfig.Predastore.BaseDir,
 		})

 		if err != nil {
@@ -363,7 +367,8 @@ var hiveStartCmd = &cobra.Command{
 		}

 		// TODO: Support ENV vars, CLI, otherwise revert to config.LoadConfig()
-		appConfig, err := config.LoadConfig(cfgFile)
+		clusterConfig, err := config.LoadConfig(cfgFile)
+		nodeConfig := clusterConfig.Nodes[clusterConfig.Node]

 		if err != nil {
 			fmt.Println("Error loading config file:", err)
@@ -375,7 +380,7 @@ var hiveStartCmd = &cobra.Command{

 		if baseDir != "" {
 			fmt.Println("Overwriting base-dir to:", baseDir)
-			appConfig.BaseDir = baseDir
+			nodeConfig.BaseDir = baseDir
 		}

 		// Overwrite defaults (CLI first, config second, env third)
@@ -383,10 +388,13 @@ var hiveStartCmd = &cobra.Command{

 		if walDir != "" {
 			fmt.Println("Overwriting wal-dir to:", walDir)
-			appConfig.WalDir = walDir
+			nodeConfig.WalDir = walDir
 		}

-		service, err := service.New("hive", appConfig)
+		// Apply changes back to cluster config
+		clusterConfig.Nodes[clusterConfig.Node] = nodeConfig
+
+		service, err := service.New("hive", clusterConfig)

 		if err != nil {
 			fmt.Println("Error starting hive service:", err)
@@ -404,7 +412,7 @@ var hiveStopCmd = &cobra.Command{
 	Run: func(cmd *cobra.Command, args []string) {
 		fmt.Println("Stopping hive service...")

-		service, err := service.New("hive", &config.Config{})
+		service, err := service.New("hive", &config.ClusterConfig{})

 		if err != nil {
 			fmt.Println("Error stopping hive service:", err)
@@ -443,7 +451,8 @@ var awsgwStartCmd = &cobra.Command{
 		fmt.Println("Loading config from:", cfgFile)

 		// TODO: Support ENV vars, CLI, otherwise revert to config.LoadConfig()
-		appConfig, err := config.LoadConfig(cfgFile)
+		clusterConfig, err := config.LoadConfig(cfgFile)
+		nodeConfig := clusterConfig.Nodes[clusterConfig.Node]

 		if err != nil {
 			fmt.Println("Error loading config file:", err)
@@ -454,30 +463,33 @@ var awsgwStartCmd = &cobra.Command{
 		awsgwHost := viper.GetString("host")
 		if awsgwHost != "" {
 			fmt.Println("Overwriting awsgw host to:", awsgwHost)
-			appConfig.AWSGW.Host = awsgwHost
+			nodeConfig.AWSGW.Host = awsgwHost
 		}

 		awsgwTlsCert := viper.GetString("tls-cert")
 		if awsgwTlsCert != "" {
 			fmt.Println("Overwriting awsgw tls-cert to:", awsgwTlsCert)
-			appConfig.AWSGW.TLSCert = awsgwTlsCert
+			nodeConfig.AWSGW.TLSCert = awsgwTlsCert
 		}

 		awsgwTlsKey := viper.GetString("tls-key")

 		if awsgwTlsKey != "" {
 			fmt.Println("Overwriting awsgw tls-key to:", awsgwTlsKey)
-			appConfig.AWSGW.TLSKey = awsgwTlsKey
+			nodeConfig.AWSGW.TLSKey = awsgwTlsKey
 		}

 		baseDir := viper.GetString("base-dir")

 		if awsgwTlsKey != "" {
 			fmt.Println("Overwriting awsgw base-dir to:", baseDir)
-			appConfig.BaseDir = baseDir
+			nodeConfig.BaseDir = baseDir
 		}

-		service, err := service.New("awsgw", appConfig)
+		// Apply changes back to cluster config
+		clusterConfig.Nodes[clusterConfig.Node] = nodeConfig
+
+		service, err := service.New("awsgw", clusterConfig)

 		if err != nil {
 			fmt.Println("Error starting awsgw service:", err)
@@ -495,7 +507,7 @@ var awsgwStopCmd = &cobra.Command{
 	Run: func(cmd *cobra.Command, args []string) {
 		fmt.Println("Stopping awsgw service...")

-		service, err := service.New("awsgw", &config.Config{})
+		service, err := service.New("awsgw", &config.ClusterConfig{})

 		if err != nil {
 			fmt.Println("Error stopping awsgw service:", err)
@@ -1,32 +1,38 @@
 # Domain configuration file in TOML format.
 version = "1.0"
-region = "{{.Region}}"
-account_id = "{{.AccountID}}"
+epoch = 1
+node = "{{.Node}}"

+[nodes.{{.Node}}]
+node = "{{.Node}}"
+region = "{{.Region}}"
+az = "{{.Az}}"
+
+account_id = "{{.AccountID}}"
 access_key = "{{.AccessKey}}"
 secret_key = "{{.SecretKey}}"

-[daemon]
-host = "0.0.0.0:4432"
+[nodes.{{.Node}}.daemon]
+host = "0.0.0.0:{{.Port}}"
 tlskey = "config/server.key"
 tlscert = "config/server.pem"

-[awsgw]
+[nodes.{{.Node}}.awsgw]
 host = "0.0.0.0:9999"
 tlskey = "config/server.key"
 tlscert = "config/server.pem"
 debug = true
 config = "awsgw/awsgw.toml"

-[nats]
+[nodes.{{.Node}}.nats]
 host = "0.0.0.0:4222"

-[nats.acl]
+[nodes.{{.Node}}.nats.acl]
 token = "{{.NatsToken}}"

-[nats.sub]
+[nodes.{{.Node}}.nats.sub]

-[predastore]
+[nodes.{{.Node}}.predastore]
 host = "0.0.0.0:8443"
 region = "{{.Region}}"
 bucket = "predastore"
@@ -11,17 +11,31 @@ import (
 	"github.com/spf13/viper"
 )

+type ClusterConfig struct {
+	Epoch   uint64            `mapstructure:"epoch"`   // bump when leader commits changes
+	Node    string            `mapstructure:"node"`    // my node name
+	Version string            `mapstructure:"version"` // hive version
+	Nodes   map[string]Config `mapstructure:"nodes"`   // full config for every node
+}
+
 // Config holds all configuration for the application
 type Config struct {
+	// Node config
+	Node    string `mapstructure:"node"`
+	Host    string `mapstructure:"host"` // Unique hostname or IP of this node
+	Region  string `mapstructure:"region"`
+	AZ      string `mapstructure:"az"`
+	DataDir string `mapstructure:"data_dir"`
+
 	Daemon     DaemonConfig     `mapstructure:"daemon"`
 	NATS       NATSConfig       `mapstructure:"nats"`
 	Predastore PredastoreConfig `mapstructure:"predastore"`
 	AWSGW      AWSGWConfig      `mapstructure:"awsgw"`

 	// Authentication
+	// TODO: Move to more appropriate setting above
 	AccessKey string `mapstructure:"access_key"`
 	SecretKey string `mapstructure:"secret_key"`
-	Host      string `mapstructure:"host"`
 	BaseDir   string `mapstructure:"base_dir"`
 	WalDir    string `mapstructure:"wal_dir"`
 }
@@ -33,7 +47,7 @@ type AWSGWConfig struct {
 	Config  string `mapstructure:"config"`

 	Debug         bool `mapstructure:"debug"`
-	ExpectedNodes int  `mapstructure:"expected_nodes"`
+	ExpectedNodes int  `mapstructure:"expected_nodes"` // TODO: Replace with root cluster config
 }

 type PredastoreConfig struct {
@@ -106,25 +120,6 @@ type EBSDeleteResponse struct {
 }

 // EC2, TODO: Move to vm.go or more applicable place
-type EC2Response struct {
-	InstanceID string
-	Hostname   string
-	Success    bool
-	Status     string
-	Error      string
-}
-
-type EC2DescribeRequest struct {
-	InstanceID string
-}
-
-type EC2DescribeResponse struct {
-	InstanceID string
-	Hostname   string
-	Status     string
-	Error      string
-}
-
 type EC2StartInstancesRequest struct {
 	InstanceID string
 }
@@ -146,45 +141,14 @@ func (ec2StartInstanceResponse EC2StartInstancesResponse) Respond(msg *nats.Msg)

 	msg.Respond(response)

-}
-func (ec2DescribeResponse EC2DescribeResponse) Respond(msg *nats.Msg) {
-
-	response, err := json.Marshal(ec2DescribeResponse)
-	if err != nil {
-		slog.Error("Failed to marshal response", "err", err)
-		return
-	}
-
-	msg.Respond(response)
-
-}
-
-func (ec2Response EC2Response) Respond(msg *nats.Msg) {
-
-	response, err := json.Marshal(ec2Response)
-	if err != nil {
-		slog.Error("Failed to marshal response", "err", err)
-		return
-	}
-
-	msg.Respond(response)
-
 }

 // LoadConfig loads the configuration from file and environment variables
-func LoadConfig(configPath string) (*Config, error) {
+func LoadConfig(configPath string) (*ClusterConfig, error) {
 	// Set environment variable prefix
 	viper.SetEnvPrefix("HIVE")
 	viper.AutomaticEnv()

-	// Set default values
-	//viper.SetDefault("host", "https://localhost:8443/")
-	//viper.SetDefault("base_dir", "/tmp/vb/")
-
-	//viper.SetDefault("daemon.host", "0.0.0.0:4432")
-	//viper.SetDefault("nats.host", "0.0.0.0:4222")
-	//viper.SetDefault("nats.sub.subject", "ec2.>")
-
 	// Try to load config file if it exists
 	if configPath != "" {
 		// Check if file exists
@@ -202,26 +166,10 @@ func LoadConfig(configPath string) (*Config, error) {
 	}

 	// Create config struct
-	var config Config
+	var config ClusterConfig
 	if err := viper.Unmarshal(&config); err != nil {
 		return nil, fmt.Errorf("error unmarshaling config: %w", err)
 	}

-	// Validate required fields (can be overwritten by env vars)
-	/*
-		if config.NATS.Host == "" {
-			return nil, fmt.Errorf("NATS host is required")
-		}
-
-		if config.AccessKey == "" {
-			return nil, fmt.Errorf("access key is required")
-		}
-		if config.SecretKey == "" {
-			return nil, fmt.Errorf("secret key is required")
-		}
-	*/
-
-	//	fmt.Println("Config: ", config)
-
 	return &config, nil
 }
@@ -71,6 +71,8 @@ type ResourceManager struct {

 // Daemon represents the main daemon service
 type Daemon struct {
+	node            string
+	clusterConfig   *config.ClusterConfig
 	config          *config.Config
 	natsConn        *nats.Conn
 	resourceMgr     *ResourceManager
@@ -178,16 +180,21 @@ func NewResourceManager() *ResourceManager {
 }

 // NewDaemon creates a new daemon instance
-func NewDaemon(cfg *config.Config) *Daemon {
+func NewDaemon(cfg *config.ClusterConfig) *Daemon {
 	ctx, cancel := context.WithCancel(context.Background())

 	// If WalDir is not set, use BaseDir
-	if cfg.WalDir == "" {
-		cfg.WalDir = cfg.BaseDir
+	config := cfg.Nodes[cfg.Node]
+	if cfg.Nodes[cfg.Node].WalDir == "" {
+		config.WalDir = config.BaseDir
+
+		cfg.Nodes[cfg.Node] = config
 	}

 	return &Daemon{
-		config:            cfg,
+		node:              cfg.Node,
+		clusterConfig:     cfg,
+		config:            &config,
 		resourceMgr:       NewResourceManager(),
 		ctx:               ctx,
 		cancel:            cancel,
@@ -488,39 +495,6 @@ func (d *Daemon) handleEC2StartInstances(msg *nats.Msg) {

 }

-func (d *Daemon) handleEC2Describe(msg *nats.Msg) {
-
-	var ec2DescribeRequest config.EC2DescribeRequest
-
-	if err := json.Unmarshal(msg.Data, &ec2DescribeRequest); err != nil {
-		log.Printf("Error unmarshaling EC2 describe request: %v", err)
-		return
-	}
-
-	slog.Info("EC2 Describe Request", "instanceId", ec2DescribeRequest.InstanceID)
-
-	var ec2DescribeResponse config.EC2DescribeResponse
-
-	// Check if the instance is running on this node
-	d.Instances.Mu.Lock()
-	defer d.Instances.Mu.Unlock()
-
-	instance, ok := d.Instances.VMS[ec2DescribeRequest.InstanceID]
-
-	if !ok {
-		slog.Error("EC2 Describe Request - Instance not found", "instanceId", ec2DescribeRequest.InstanceID)
-		ec2DescribeResponse.InstanceID = ec2DescribeRequest.InstanceID
-		ec2DescribeResponse.Error = fmt.Sprintf("Instance %s not found", ec2DescribeRequest.InstanceID)
-		ec2DescribeResponse.Respond(msg)
-		return
-	}
-
-	ec2DescribeResponse.InstanceID = instance.ID
-	ec2DescribeResponse.Status = instance.Status
-	ec2DescribeResponse.Respond(msg)
-
-}
-
 func (d *Daemon) SendQMPCommand(q *qmp.QMPClient, cmd qmp.QMPCommand, instanceId string) (*qmp.QMPResponse, error) {

 	// Confirm QMP client is initialized
@@ -55,6 +55,12 @@ func createTestDaemon(t *testing.T, natsURL string) *Daemon {
 		os.RemoveAll(tmpDir)
 	})

+	// New cluster config
+	clusterCfg := &config.ClusterConfig{
+		Node:  "node-1",
+		Nodes: map[string]config.Config{},
+	}
+
 	cfg := &config.Config{
 		BaseDir: tmpDir,
 		WalDir:  tmpDir,
@@ -76,7 +82,9 @@ func createTestDaemon(t *testing.T, natsURL string) *Daemon {
 		SecretKey: "test-secret-key",
 	}

-	daemon := NewDaemon(cfg)
+	clusterCfg.Nodes["node-1"] = *cfg
+
+	daemon := NewDaemon(clusterCfg)

 	// Connect to NATS
 	nc, err := nats.Connect(natsURL)
@@ -352,6 +360,12 @@ func TestDaemon_Initialization(t *testing.T) {
 	require.NoError(t, err)
 	defer os.RemoveAll(tmpDir)

+	// New cluster config
+	clusterCfg := &config.ClusterConfig{
+		Node:  "node-1",
+		Nodes: map[string]config.Config{},
+	}
+
 	cfg := &config.Config{
 		BaseDir: tmpDir,
 		WalDir:  tmpDir,
@@ -362,7 +376,9 @@ func TestDaemon_Initialization(t *testing.T) {
 		SecretKey: "test-secret",
 	}

-	daemon := NewDaemon(cfg)
+	clusterCfg.Nodes["node-1"] = *cfg
+
+	daemon := NewDaemon(clusterCfg)

 	assert.NotNil(t, daemon)
 	assert.NotNil(t, daemon.resourceMgr)
@@ -17,12 +17,12 @@ import (
 var serviceName = "awsgw"

 type Service struct {
-	Config *config.Config
+	Config *config.ClusterConfig
 }

 func New(cfg interface{}) (svc *Service, err error) {
 	svc = &Service{
-		Config: cfg.(*config.Config),
+		Config: cfg.(*config.ClusterConfig),
 	}
 	return svc, nil
 }
@@ -54,11 +54,13 @@ func (svc *Service) Reload() (err error) {
 	return nil
 }

-func launchService(config *config.Config) (err error) {
+func launchService(config *config.ClusterConfig) (err error) {
+
+	nodeConfig := config.Nodes[config.Node]

 	// Connect to NATS for service communication
 	opts := []nats.Option{
-		nats.Token(config.NATS.ACL.Token),
+		nats.Token(nodeConfig.NATS.ACL.Token),
 		nats.ReconnectWait(time.Second),
 		nats.MaxReconnects(-1), // Infinite reconnects
 		nats.DisconnectErrHandler(func(nc *nats.Conn, err error) {
@@ -69,26 +71,26 @@ func launchService(config *config.Config) (err error) {
 		}),
 	}

-	natsConn, err := nats.Connect(config.NATS.Host, opts...)
+	natsConn, err := nats.Connect(nodeConfig.NATS.Host, opts...)
 	if err != nil {
 		slog.Error("Failed to connect to NATS", "err", err)
 		return err
 	}
 	defer natsConn.Close()

-	slog.Info("Connected to NATS server", "host", config.NATS.Host)
+	slog.Info("Connected to NATS server", "host", nodeConfig.NATS.Host)

 	// Append Base dir if config has no leading path
-	if config.BaseDir != "" && !strings.HasPrefix(config.AWSGW.Config, "/") {
-		config.AWSGW.Config = fmt.Sprintf("%s/%s", config.BaseDir, config.AWSGW.Config)
+	if nodeConfig.BaseDir != "" && !strings.HasPrefix(nodeConfig.AWSGW.Config, "/") {
+		nodeConfig.AWSGW.Config = fmt.Sprintf("%s/%s", nodeConfig.BaseDir, nodeConfig.AWSGW.Config)
 	}

 	// Create gateway with NATS connection
 	gw := gateway.GatewayConfig{
-		Debug:          config.AWSGW.Debug,
+		Debug:          nodeConfig.AWSGW.Debug,
 		DisableLogging: false,
 		NATSConn:       natsConn,
-		Config:         config.AWSGW.Config,
+		Config:         nodeConfig.AWSGW.Config,
 	}

 	app := gw.SetupRoutes()
@@ -98,7 +100,7 @@ func launchService(config *config.Config) (err error) {
 		return err
 	}

-	log.Fatal(app.ListenTLS(config.AWSGW.Host, config.AWSGW.TLSCert, config.AWSGW.TLSKey))
+	log.Fatal(app.ListenTLS(nodeConfig.AWSGW.Host, nodeConfig.AWSGW.TLSCert, nodeConfig.AWSGW.TLSKey))

 	return nil

@@ -12,12 +12,12 @@ import (
 var serviceName = "hive"

 type Service struct {
-	Config *config.Config
+	Config *config.ClusterConfig
 }

 func New(cfg interface{}) (svc *Service, err error) {
 	svc = &Service{
-		Config: cfg.(*config.Config),
+		Config: cfg.(*config.ClusterConfig),
 	}
 	return svc, nil
 }
@@ -49,7 +49,7 @@ func (svc *Service) Reload() (err error) {
 	return nil
 }

-func launchService(config *config.Config) (err error) {
+func launchService(config *config.ClusterConfig) (err error) {

 	d := daemon.NewDaemon(config)
 	slog.Info("Starting Hive daemon ...")