[rabbitmq-discuss] AMQP, RabbitMQ and message integrity
james.gardner at noaa.gov
Thu Mar 21 17:51:12 GMT 2013
Sounds fine, thanks Emile.
On 03/21/2013 12:36 PM, Emile Joubert wrote:
> Hi James,
> On 21/03/13 17:03, James Gardner wrote:
>> Perhaps that question can largely be rephrased as what guarantees does
>> *AMQP* make about message integrity, although how messages are stored
>> and transported within RabbitMQ would surely be a part of it too.
> Yes, a hash in a header will work. The broker does not perform any
> integrity test on messages (though it will notice corruption that causes
> the length to change).
> In theory it is possible for highly specific failure modes to cause
> corruption of messages. In practice there are network checksums and
> parity checks in hardware for protection. The broker is far more likely
> to crash than deliver corrupt data unwittingly, but the hash you suggest
> would be another line of defence against this remote possibility.
More information about the rabbitmq-discuss