Fix socket name too long

Author: sjmiller609

lib/hypervisor/cloudhypervisor/process.go

@@ -9,6 +9,10 @@ import (
 	"github.com/onkernel/hypeman/lib/vmm"
 )
 
+func init() {
+	hypervisor.RegisterSocketName(hypervisor.TypeCloudHypervisor, "ch.sock")
+}
+
 // ProcessManager implements hypervisor.ProcessManager for Cloud Hypervisor.
 type ProcessManager struct{}
 
@@ -20,6 +24,11 @@ func NewProcessManager() *ProcessManager {
 // Verify ProcessManager implements the interface
 var _ hypervisor.ProcessManager = (*ProcessManager)(nil)
 
+// SocketName returns the socket filename for Cloud Hypervisor.
+func (p *ProcessManager) SocketName() string {
+	return "ch.sock"
+}
+
 // StartProcess launches a Cloud Hypervisor VMM process.
 func (p *ProcessManager) StartProcess(ctx context.Context, paths *paths.Paths, version string, socketPath string) (int, error) {
 	chVersion := vmm.CHVersion(version)

lib/hypervisor/hypervisor.go

@@ -19,6 +19,25 @@ const (
 	// Future: TypeQEMU Type = "qemu"
 )
 
+// socketNames maps hypervisor types to their socket filenames.
+// Registered by each hypervisor package's init() function.
+var socketNames = make(map[Type]string)
+
+// RegisterSocketName registers the socket filename for a hypervisor type.
+// Called by each hypervisor implementation's init() function.
+func RegisterSocketName(t Type, name string) {
+	socketNames[t] = name
+}
+
+// SocketNameForType returns the socket filename for a hypervisor type.
+// Falls back to type + ".sock" if not registered.
+func SocketNameForType(t Type) string {
+	if name, ok := socketNames[t]; ok {
+		return name
+	}
+	return string(t) + ".sock"
+}
+
 // Hypervisor defines the interface for VM management operations.
 // All hypervisor implementations must implement this interface.
 type Hypervisor interface {
@@ -92,6 +111,10 @@ type Capabilities struct {
 // This is separate from the Hypervisor interface because process management
 // happens before/after the VMM socket is available.
 type ProcessManager interface {
+	// SocketName returns the socket filename for this hypervisor.
+	// Uses short names to stay within Unix socket path length limits (SUN_LEN ~108 bytes).
+	SocketName() string
+
 	// StartProcess launches the hypervisor process.
 	// Returns the process ID of the started hypervisor.
 	StartProcess(ctx context.Context, p *paths.Paths, version string, socketPath string) (pid int, err error)

lib/instances/README.md

@@ -40,7 +40,7 @@ Manages VM instance lifecycle using Cloud Hypervisor.
       metadata.json             # State, versions, timestamps
       overlay.raw               # 50GB sparse writable overlay
       config.erofs              # Compressed config disk
-      cloud-hypervisor.sock     # Hypervisor API socket (named after hypervisor type)
+      ch.sock                   # Hypervisor API socket (abbreviated for SUN_LEN limit)
       logs/
         app.log                 # Guest application log (serial console output)
         vmm.log                 # Hypervisor log (stdout+stderr)

lib/instances/create.go

@@ -202,7 +202,15 @@ func (m *manager) createInstance(
 	// 8. Get default kernel version
 	kernelVer := m.systemManager.GetDefaultKernelVersion()
 
-	// 9. Validate, resolve, and auto-bind devices (GPU passthrough)
+	// 9. Get process manager for hypervisor type (needed for socket name)
+	hvType := hypervisor.TypeCloudHypervisor
+	pm, err := m.getProcessManager(hvType)
+	if err != nil {
+		log.ErrorContext(ctx, "failed to get process manager", "error", err)
+		return nil, fmt.Errorf("get process manager: %w", err)
+	}
+
+	// 10. Validate, resolve, and auto-bind devices (GPU passthrough)
 	// Track devices we've marked as attached for cleanup on error.
 	// The cleanup closure captures this slice by reference, so it will see
 	// whatever devices have been attached when cleanup runs.
@@ -256,7 +264,7 @@ func (m *manager) createInstance(
 		log.DebugContext(ctx, "validated devices for passthrough", "id", id, "devices", resolvedDeviceIDs)
 	}
 
-	// 10. Create instance metadata
+	// 11. Create instance metadata
 	stored := &StoredMetadata{
 		Id:                id,
 		Name:              req.Name,
@@ -271,30 +279,30 @@ func (m *manager) createInstance(
 		StartedAt:         nil,
 		StoppedAt:         nil,
 		KernelVersion:     string(kernelVer),
-		HypervisorType:    hypervisor.TypeCloudHypervisor,
+		HypervisorType:    hvType,
 		HypervisorVersion: string(vmm.V49_0), // Use latest
-		SocketPath:        m.paths.InstanceSocket(id, string(hypervisor.TypeCloudHypervisor)),
+		SocketPath:        m.paths.InstanceSocket(id, pm.SocketName()),
 		DataDir:           m.paths.InstanceDir(id),
 		VsockCID:          vsockCID,
 		VsockSocket:       vsockSocket,
 		Devices:           resolvedDeviceIDs,
 	}
 
-	// 11. Ensure directories
+	// 12. Ensure directories
 	log.DebugContext(ctx, "creating instance directories", "instance_id", id)
 	if err := m.ensureDirectories(id); err != nil {
 		log.ErrorContext(ctx, "failed to create directories", "instance_id", id, "error", err)
 		return nil, fmt.Errorf("ensure directories: %w", err)
 	}
 
-	// 12. Create overlay disk with specified size
+	// 13. Create overlay disk with specified size
 	log.DebugContext(ctx, "creating overlay disk", "instance_id", id, "size_bytes", stored.OverlaySize)
 	if err := m.createOverlayDisk(id, stored.OverlaySize); err != nil {
 		log.ErrorContext(ctx, "failed to create overlay disk", "instance_id", id, "error", err)
 		return nil, fmt.Errorf("create overlay disk: %w", err)
 	}
 
-	// 13. Allocate network (if network enabled)
+	// 14. Allocate network (if network enabled)
 	var netConfig *network.NetworkConfig
 	if networkName != "" {
 		log.DebugContext(ctx, "allocating network", "instance_id", id, "network", networkName)
@@ -319,7 +327,7 @@ func (m *manager) createInstance(
 		})
 	}
 
-	// 14. Validate and attach volumes
+	// 15. Validate and attach volumes
 	if len(req.Volumes) > 0 {
 		log.DebugContext(ctx, "validating volumes", "instance_id", id, "count", len(req.Volumes))
 		for _, volAttach := range req.Volumes {
@@ -359,30 +367,30 @@ func (m *manager) createInstance(
 		stored.Volumes = req.Volumes
 	}
 
-	// 15. Create config disk (needs Instance for buildVMConfig)
+	// 16. Create config disk (needs Instance for buildVMConfig)
 	inst := &Instance{StoredMetadata: *stored}
 	log.DebugContext(ctx, "creating config disk", "instance_id", id)
 	if err := m.createConfigDisk(ctx, inst, imageInfo, netConfig); err != nil {
 		log.ErrorContext(ctx, "failed to create config disk", "instance_id", id, "error", err)
 		return nil, fmt.Errorf("create config disk: %w", err)
 	}
 
-	// 16. Save metadata
+	// 17. Save metadata
 	log.DebugContext(ctx, "saving instance metadata", "instance_id", id)
 	meta := &metadata{StoredMetadata: *stored}
 	if err := m.saveMetadata(meta); err != nil {
 		log.ErrorContext(ctx, "failed to save metadata", "instance_id", id, "error", err)
 		return nil, fmt.Errorf("save metadata: %w", err)
 	}
 
-	// 17. Start VMM and boot VM
+	// 18. Start VMM and boot VM
 	log.InfoContext(ctx, "starting VMM and booting VM", "instance_id", id)
 	if err := m.startAndBootVM(ctx, stored, imageInfo, netConfig); err != nil {
 		log.ErrorContext(ctx, "failed to start and boot VM", "instance_id", id, "error", err)
 		return nil, err
 	}
 
-	// 18. Update timestamp after VM is running
+	// 19. Update timestamp after VM is running
 	now := time.Now()
 	stored.StartedAt = &now
 

lib/instances/storage.go

@@ -11,16 +11,16 @@ import (
 
 // Filesystem structure:
 // {dataDir}/guests/{instance-id}/
-//   metadata.json           # Instance metadata
-//   overlay.raw             # Configurable sparse overlay disk (default 10GB)
-//   config.ext4             # Read-only config disk (generated)
-//   cloud-hypervisor.sock   # Hypervisor API socket (named after hypervisor type)
+//   metadata.json      # Instance metadata
+//   overlay.raw        # Configurable sparse overlay disk (default 10GB)
+//   config.ext4        # Read-only config disk (generated)
+//   ch.sock            # Hypervisor API socket (abbreviated name for SUN_LEN limit)
 //   logs/
-//     app.log               # Guest application log (serial console output)
-//     vmm.log               # Hypervisor log (stdout+stderr combined)
-//     hypeman.log           # Hypeman operations log (actions taken on this instance)
+//     app.log          # Guest application log (serial console output)
+//     vmm.log          # Hypervisor log (stdout+stderr combined)
+//     hypeman.log      # Hypeman operations log (actions taken on this instance)
 //   snapshots/
-//     snapshot-latest/      # Snapshot directory
+//     snapshot-latest/ # Snapshot directory
 //       config.json
 //       memory-ranges
 

lib/network/derive.go

@@ -7,6 +7,7 @@ import (
 	"net"
 	"os"
 
+	"github.com/onkernel/hypeman/lib/hypervisor"
 	"github.com/onkernel/hypeman/lib/logger"
 	"github.com/onkernel/hypeman/lib/vmm"
 )
@@ -49,7 +50,7 @@ func (m *manager) deriveAllocation(ctx context.Context, instanceID string) (*All
 	netmask := fmt.Sprintf("%d.%d.%d.%d", ipNet.Mask[0], ipNet.Mask[1], ipNet.Mask[2], ipNet.Mask[3])
 
 	// 4. Try to derive from running VM first
-	socketPath := m.paths.InstanceSocket(instanceID, meta.HypervisorType)
+	socketPath := m.paths.InstanceSocket(instanceID, hypervisor.SocketNameForType(hypervisor.Type(meta.HypervisorType)))
 	if fileExists(socketPath) {
 		client, err := vmm.NewVMM(socketPath)
 		if err == nil {

lib/paths/paths.go

@@ -144,9 +144,11 @@ func (p *Paths) InstanceVolumeOverlaysDir(instanceID string) string {
 	return filepath.Join(p.InstanceDir(instanceID), "vol-overlays")
 }
 
-// InstanceSocket returns the path to instance API socket for the given hypervisor type.
-func (p *Paths) InstanceSocket(id string, hypervisorType string) string {
-	return filepath.Join(p.InstanceDir(id), hypervisorType+".sock")
+// InstanceSocket returns the path to instance API socket.
+// The socketName should be obtained from hypervisor.Type.SocketName() to ensure
+// it stays within Unix socket path length limits (SUN_LEN ~108 bytes).
+func (p *Paths) InstanceSocket(id string, socketName string) string {
+	return filepath.Join(p.InstanceDir(id), socketName)
 }
 
 // InstanceVsockSocket returns the path to instance vsock socket.

lib/vmm/client.go

@@ -152,6 +152,11 @@ func StartProcessWithArgs(ctx context.Context, p *paths.Paths, version CHVersion
 	defer cancel()
 
 	if err := waitForSocket(waitCtx, socketPath, 5*time.Second); err != nil {
+		// Read vmm.log to understand why socket wasn't created
+		vmmLogPath := filepath.Join(logsDir, "vmm.log")
+		if logData, readErr := os.ReadFile(vmmLogPath); readErr == nil && len(logData) > 0 {
+			return 0, fmt.Errorf("%w; vmm.log: %s", err, string(logData))
+		}
 		return 0, err
 	}