Spring AMQP：CorrelationId在发送和接收消息的时刻之间发生变化

Question

While I was testing the send and receive methods which I created for my project I ran into a strange problem. When I send a certain message using a correlationId that is based on a UUID object, the receiving party gets a slightly modified version of this correlationId (which cannot be deserialised).

On the sending side I do this:

MessageProperties properties = new MessageProperties();
properties.setCorrelationId(MessageSerializer.serialize(UUID.randomUUID().toString()));

On my last test the UUID generated was: " d4170243-9e7e-4c42-9168-f9da4debc5bb "

This generates the following correlationId (when serialized):

[-84, -19, 0, 5, 116, 0, 36, 100, 52, 49, 55, 48, 50, 52, 51, 45, 57, 101, 55, 101, 45, 52, 99, 52, 50, 45, 57, 49, 54, 56, 45, 102, 57, 100, 97, 52, 100, 101, 98, 99, 53, 98, 98]

When I receive the message on the other side this id is slightly changed:

[-17, -65, -67, -17, -65, -67, 0, 5, 116, 0, 36, 100, 52, 49, 55, 48, 50, 52, 51, 45, 57, 101, 55, 101, 45, 52, 99, 52, 50, 45, 57, 49, 54, 56, 45, 102, 57, 100, 97, 52, 100, 101, 98, 99, 53, 98, 98]

When using the RabbitMQ management plugin I noticed that the id already changed upon arrival at the queue.

Tracing my code on the sending side brings me to the send option of the RabbitTemplate class.

RabbitTemplate template = new RabbitTemplate(connection);
template.setExchange("amq.direct");
template.setRoutingKey("some.route");
template.send(message);

But I can't figure out what's causing this problem. I guess it's just me using the correlationId option the wrong way. Could someone help me out?

Appreciate it.

Answer 1

Explanation is the following:

1. You serialize the UUID string to a byte array
2. Your serialization prepends non ascii character to this array ( [-17, -65, -67, -17, -65, -67, 0, 5, 116, 0, 36,...] )
3. The reference documentation states that the correlation id is a shortstr. RabbitMQ client converts this byte array to a string using new String(yourArray , "UTF-8") .
4. The non ascii character "corrupt" the conversion from byte[] to string

You can get the same result with the following code:

new String(MessageSerializer.serialize(UUID.randomUUID().toString()) , "UTF-8").getByte("UTF-8")

Which will return:

[-17, -65, -67, -17, -65, -67, 0, 5, 116, 0, 36, 100, 52, 49, 55, 48, 50, 52, 51, 45, 57, 101, 55, 101, 45, 52, 99, 52, 50, 45, 57, 49, 54, 56, 45, 102, 57, 100, 97, 52, 100, 101, 98, 99, 53, 98, 98]