diff --git a/internal/machine/corroservice/service.go b/internal/machine/corroservice/service.go
index cf24bfee..6c0360ea 100644
--- a/internal/machine/corroservice/service.go
+++ b/internal/machine/corroservice/service.go
@@ -2,4 +2,6 @@ package corroservice
 
 type Service interface {
 	Start() error
+	Restart() error
+	Running() bool
 }
diff --git a/internal/machine/corroservice/systemd.go b/internal/machine/corroservice/systemd.go
index c520df06..fb10185e 100644
--- a/internal/machine/corroservice/systemd.go
+++ b/internal/machine/corroservice/systemd.go
@@ -11,6 +11,7 @@ const DefaultSystemdUnit = "uncloud-corrosion.service"
 type SystemdService struct {
 	DataDir string
 	Unit    string
+	running bool
 }
 
 func DefaultSystemdService(dataDir string) *SystemdService {
@@ -25,5 +26,23 @@ func (s *SystemdService) Start() error {
 		return fmt.Errorf("systemctl start %s: %w", s.Unit, err)
 	}
 	slog.Info("Corrosion systemd service started.", "unit", s.Unit)
+
+	// TODO: run a goroutine to check the status of the service and log any errors in the uncloud log.
+	s.running = true
 	return nil
 }
+
+func (s *SystemdService) Restart() error {
+	if _, err := exec.Command("systemctl", "restart", s.Unit).Output(); err != nil {
+		return fmt.Errorf("systemctl restart %s: %w", s.Unit, err)
+	}
+	slog.Info("Corrosion systemd service restarted.", "unit", s.Unit)
+
+	// TODO: run a goroutine to check the status of the service and log any errors in the uncloud log.
+	s.running = true
+	return nil
+}
+
+func (s *SystemdService) Running() bool {
+	return s.running
+}
diff --git a/internal/machine/machine.go b/internal/machine/machine.go
index b6069524..be642ec2 100644
--- a/internal/machine/machine.go
+++ b/internal/machine/machine.go
@@ -192,6 +192,24 @@ func (m *Machine) Run(ctx context.Context) error {
 	)
 	close(m.started)
 
+	// Configure and start the corrosion service on the loopback if the machine is not initialised as a cluster
+	// member. This provides the store required for the machine to initialise a new cluster on it.
+	if !m.IsInitialised() {
+		// Needs to run in a goroutine because the corrosion systemd service depends on the readiness of the daemon
+		// indicated by closing the started channel.
+		errGroup.Go(func() error {
+			if err := m.configureCorrosion(); err != nil {
+				return fmt.Errorf("configure corrosion service: %w", err)
+			}
+			slog.Info("Configured corrosion service.", "dir", m.config.CorrosionDir)
+
+			if err := m.config.CorrosionService.Start(); err != nil {
+				return fmt.Errorf("start corrosion service: %w", err)
+			}
+			return nil
+		})
+	}
+
 	// Control loop for managing the network controller.
 	errGroup.Go(
 		func() error {
@@ -212,13 +230,15 @@ func (m *Machine) Run(ctx context.Context) error {
 				// It can be reset when leaving the cluster and then re-initialised again with a new configuration.
 				case <-m.initialised:
 					var err error
-					slog.Info("Starting network controller.")
-					networkServer := newGRPCServer(m, m.cluster)
-
+					// Ensure the corrosion config is up to date, including a new gossip address if the machine
+					// has just joined a cluster.
 					if err = m.configureCorrosion(); err != nil {
 						return fmt.Errorf("configure corrosion service: %w", err)
 					}
+					slog.Info("Configured corrosion service.", "dir", m.config.CorrosionDir)
 
+					slog.Info("Starting network controller.")
+					networkServer := newGRPCServer(m, m.cluster)
 					ctrl, err = newNetworkController(m.state, networkServer, m.config.CorrosionService, m.newMachinesCh)
 					if err != nil {
 						return fmt.Errorf("initialise network controller: %w", err)
@@ -296,6 +316,12 @@ func (m *Machine) configureCorrosion() error {
 	configPath := filepath.Join(m.config.CorrosionDir, "config.toml")
 	schemaPath := filepath.Join(m.config.CorrosionDir, "schema.sql")
 
+	// Use a loopback address as the gossip address (required) unless the machine has joined a cluster
+	// and has a management IP.
+	gossipAddr := netip.AddrPortFrom(netip.AddrFrom4([4]byte{127, 0, 0, 1}), corroservice.DefaultGossipPort)
+	if m.state.Network.ManagementIP.IsValid() {
+		gossipAddr = netip.AddrPortFrom(m.state.Network.ManagementIP, corroservice.DefaultGossipPort)
+	}
 	// TODO: use a partial list of machine peers for bootstrapping if the cluster is large.
 	var bootstrap []string
 	for _, peer := range m.state.Network.Peers {
@@ -311,7 +337,7 @@ func (m *Machine) configureCorrosion() error {
 			SchemaPaths: []string{schemaPath},
 		},
 		Gossip: corroservice.GossipConfig{
-			Addr:      netip.AddrPortFrom(m.state.Network.ManagementIP, corroservice.DefaultGossipPort),
+			Addr:      gossipAddr,
 			Bootstrap: bootstrap,
 			Plaintext: true,
 		},
diff --git a/internal/machine/network.go b/internal/machine/network.go
index e19d2653..5e37b464 100644
--- a/internal/machine/network.go
+++ b/internal/machine/network.go
@@ -56,8 +56,15 @@ func (nc *networkController) Run(ctx context.Context) error {
 	}
 	slog.Info("WireGuard network configured.")
 
-	if err := nc.corroService.Start(); err != nil {
-		return err
+	if nc.corroService.Running() {
+		// Corrosion service was running before the WireGuard network was configured so we need to restart it.
+		if err := nc.corroService.Restart(); err != nil {
+			return fmt.Errorf("restart corrosion service: %w", err)
+		}
+	} else {
+		if err := nc.corroService.Start(); err != nil {
+			return fmt.Errorf("start corrosion service: %w", err)
+		}
 	}
 	// TODO: Figure out if we need to manually stop the corrosion service when the context is done or just
 	//  rely on systemd to handle service dependencies on its own .