最终找到了解决办法。编码不正确--内置的writebytes（）方法会自动为您写入长度。
然后在consumer中，我们必须通过genericdatumwriter，写入二进制流，然后用reflectdatumreader从二进制流中读取。这将自动调用uuiasbytesencoding read（）方法并反序列化uuid。
我的消费者看起来像这样（作为消费者组executor服务演练的一部分）：

/**
 * Start a single consumer instance
 * This will use the schema built into the IndexedRecord to decode and create key/value for the message
 */
public void run() {
    ConsumerIterator it = this.stream.iterator();
    while (it.hasNext()) {
        MessageAndMetadata messageAndMetadata = it.next();
        try {
            String key = (String) messageAndMetadata.key();
            IndexedRecord value = (IndexedRecord) messageAndMetadata.message();

            ByteArrayOutputStream bytes = new ByteArrayOutputStream();

            GenericDatumWriter<Object> genericRecordWriter = new GenericDatumWriter<>(value.getSchema());
            genericRecordWriter.write(value, EncoderFactory.get().directBinaryEncoder(bytes, null));

            ReflectDatumReader<T> reflectDatumReader = new ReflectDatumReader<>(value.getSchema());
            T newObject = reflectDatumReader.read(null, DecoderFactory.get().binaryDecoder(bytes.toByteArray(), null));
            IOUtils.closeQuietly(bytes);

            System.out.println("************CONSUMED:  " + key + ": "+ newObject);

        } catch(SerializationException e) {
            e.printStackTrace();
        } catch (IOException e) {
            e.printStackTrace();
        }
    }

    System.out.println("Shutting down Thread: " + this.threadNumber);
}

那么新的uuidasbytesen编码将如下所示：

public class UUIDAsBytesEncoding extends CustomEncoding<UUID> {

    public UUIDAsBytesEncoding() {
        List<Schema> union = Arrays.asList(Schema.create(Schema.Type.NULL), Schema.create(Schema.Type.BYTES));
        union.get(1).addProp("CustomEncoding", "UUIDAsBytesEncoding");

        schema = Schema.createUnion(union);
    }

    @Override
    protected void write(Object datum, Encoder out) throws IOException {
        if(datum != null) {
            // encode the position of the data in the union
            out.writeLong(1);

            // convert uuid to bytes
            byte[] bytes = new byte[16];
            Conversion.uuidToByteArray(((UUID) datum), bytes, 0, 16);

            // write the data
            out.writeBytes(bytes);
        } else {
            // position of null in union
            out.writeLong(0);
        }
    }

    @Override
    protected UUID read(Object reuse, Decoder in) throws IOException {
        // get index in union
        int index = in.readIndex();
        if (index == 1) {
            // read in 16 bytes of data
            ByteBuffer b = ByteBuffer.allocate(16);
            in.readBytes(b);

            // convert
            UUID uuid = Conversion.byteArrayToUuid(b.array(), 0);

            return uuid;
        } else {
            // no uuid present
            return null;
        }
    }
}

这也是一个如何实现customencoding avro类的示例。avro的当前版本没有内置uuid序列化程序，因此这是解决该问题的一种方法。