another cc-connect instance is already running (PID %d) with

Error message

another cc-connect instance is already running (PID %d) with config %s

What it means

Windows variant of lock contention: syscall.CreateFile on the lock file failed and readPIDFromLockFile recovered a positive PID, so this error names the running instance's PID and config path. CreateFile fails when another instance holds the lock with no share/write access, or on permission problems — but the intent of the message is "another instance is running".

Source

Thrown at cmd/cc-connect/instance_lock_windows.go:52

	pathPtr, err := syscall.UTF16PtrFromString(lockPath)
	if err != nil {
		return nil, fmt.Errorf("cannot convert lock path: %w", err)
	}

	handle, createErr := syscall.CreateFile(
		pathPtr,
		syscall.GENERIC_READ|syscall.GENERIC_WRITE,
		syscall.FILE_SHARE_READ,
		nil,
		syscall.OPEN_ALWAYS,
		syscall.FILE_ATTRIBUTE_NORMAL,
		0,
	)

	if createErr != nil {
		pid := readPIDFromLockFile(lockPath)
		if pid > 0 {
			return nil, fmt.Errorf("another cc-connect instance is already running (PID %d) with config %s", pid, configPath)
		}
		return nil, fmt.Errorf("another cc-connect instance is already running with config %s", configPath)
	}

	pid := os.Getpid()
	syscall.SetFilePointer(handle, 0, nil, syscall.FILE_BEGIN)
	syscall.SetEndOfFile(handle)
	var written uint32
	syscall.WriteFile(handle, []byte(fmt.Sprintf("%d\n", pid)), &written, nil)
	syscall.FlushFileBuffers(handle)

	return &InstanceLock{
		handle:   handle,
		path:     lockPath,
		acquired: true,
	}, nil
}

View on GitHub (pinned to 4000b2338a)

Solutions

Confirm the PID is live (tasklist /FI "PID eq <PID>") and stop the existing instance (taskkill /PID <PID> or stop the service).
Use a different --config path if a genuinely independent instance is required.
If the PID is stale, delete the lock file and retry.

Defensive patterns

Strategy: try-catch

Try / catch

lock, err := AcquireInstanceLock(configPath)
if err != nil {
    var pid int
    if _, scanErr := fmt.Sscanf(err.Error(), "another cc-connect instance is already running (PID %d)", &pid); scanErr == nil {
        fmt.Fprintf(os.Stderr, "stop PID %d first.\n", pid)
    }
    os.Exit(1)
}

Prevention

Run instances via Windows services or scheduled tasks that enforce singletons.
Always use the same user account for the daemon.
Use distinct config paths for parallel instances.

When it happens

Trigger: AcquireInstanceLock on Windows while another cc-connect process holds an open handle (created without FILE_SHARE_WRITE) on <configDir>\. <configBase>.lock and its PID was read back successfully.

Common situations: Second launch while the service/scheduled-task instance is running; running under a different user while the first instance's handle denies sharing.

Related errors

AI-assisted analysis of chenhg5/cc-connect@4000b2338a (2026-09-06). Data as JSON: /api/errors/bdef8c3879cd01a0. Report an issue: GitHub.