return null;
    }

    private String getError(File touchFile, String key) {
        Properties props = read(touchFile);
        if (props != null) {
            return props.getProperty(key + ERROR_KEY_SUFFIX);
        }
        return null;
    }

    private Properties read(File touchfile) {
        if (!touchfile.canRead()) {

        buffer[bufferIndex + 36] = 0x00; // reserved

        int bytesRead = response.readParameterWordsWireFormat(buffer, bufferIndex);

        // Verify the correct number of bytes were read (37 bytes total)
        assertEquals(37, bytesRead);

        // Verify values through reflection
        Field totalParamField = response.getClass().getSuperclass().getSuperclass().getDeclaredField("totalParameterCount");

        byte[] smallOutputBuffer = new byte[5];
        TransCallNamedPipeResponse smallResponse = new TransCallNamedPipeResponse(mockConfig, smallOutputBuffer);

        byte[] buffer = new byte[100];

        // Try to read more data than output buffer can hold
        SMBProtocolDecodingException exception =
                assertThrows(SMBProtocolDecodingException.class, () -> smallResponse.readDataWireFormat(buffer, 0, 10));

The same models are shared among requests, so, it's not one model per request, or one per user or something similar.

Let's imagine that loading the model can **take quite some time**, because it has to read a lot of **data from disk**. So you don't want to do it for every request.

    /**
     * RDMA transform capabilities flag indicating response is requested
     */
    public static final int SMB_DIRECT_RESPONSE_REQUESTED = 0x00000001;

    /**
     * Default maximum size for RDMA read/write operations (1MB)
     */
    public static final int DEFAULT_RDMA_READ_WRITE_SIZE = 1048576; // 1MB

    /**
     * Default maximum number of receive credits
     */

        // Decode
        int bytesDecoded = smbInfoAllocation.decode(buffer, 0, buffer.length);

        // Verify we read exactly 20 bytes (4 + 4 + 4 + 4 + 4)
        // Note: bytesPerSect is read as Int2 but 4 bytes are consumed (padding)
        assertEquals(20, bytesDecoded);
    }

    @Test
    @DisplayName("Test capacity calculation overflow handling")

     * ServerMessageBlock would have to perform the same read
     * operation on the input stream, we might as will pull that
     * common functionality into the superclass and read wordCount
     * and byteCount worth of data.
     *
     * We will still use the readXxxWireFormat return values to
     * indicate how many bytes(note: readParameterWordsWireFormat
     * returns bytes read and not the number of words(but the

    * Esto es especialmente útil si ejecutas `await myfile.read()` una vez y luego necesitas leer el contenido nuevamente.
* `close()`: Cierra el archivo.

Como todos estos métodos son métodos `async`, necesitas "await" para ellos.

Por ejemplo, dentro de una *path operation function* `async` puedes obtener los contenidos con:

```Python
contents = await myfile.read()
```

        // Read the data
        int bytesRead = response.readParameterWordsWireFormat(buffer, bufferIndex);

        // Verify results
        assertEquals(20, bytesRead);
        assertEquals(fileAttributes, response.getAttributes());
        assertEquals(fileSize, response.getSize());

        // Verify time was read correctly (accounting for UTime conversion)

| `minio_node_drive_reads_per_sec`     | Reads per second on a drive.                             |
| `minio_node_drive_reads_kb_per_sec`  | Kilobytes read per second on a drive.                    |
| `minio_node_drive_reads_await`       | Average time for read requests to be served on a drive.  |