[rabbitmq-discuss] Journaling for RabbitMQ?
lgreg.meredith at gmail.com
Thu May 26 19:55:20 BST 2011
i was working with a client recently and they asked whether RabbitMQ had a
journaling feature. The client was used to working with MSMQ which has such
a feature and allows them to get a view of queue usage history, such as
message deposited to queue, message read from queue, etc. The question came
up in the context of debugging a message processing code path where we
couldn't readily tell whether the message had been deposited to the queue
and was being nabbed by a dispatcher faster than we could see or whether the
message was never arriving at the queue at all. The client observed that
with a journaling feature there would be no need to introduce more debugging
statements into application code. So, i'm passing their question along to
In this connection i will raise a point i've made before to this community.
With such a journaling log i can take off-the-shelf tools and probe message
traces for correctness properties. Specifically, i have already demonstrated
with a little project called XTrace that with an xml-based message history
format we can use schema conformance to probe message trace for properties.
The basic idea is to encode a property as a schema. The beauty of this
approach is that it is post facto and does not require a specific schema to
govern message trace history. Rather, an application programmer devises
schema to investigate the message trace history to look for conformance to
protocol or validate a guess about where/how things might be going awry.
i mention this latter point in connection with the client's question to
amplify the argument for the feature. Their sense of the separation of
concerns and independent value for RabbitMQ is right on track and only the
tip of the iceberg of what can be done.
7329 39th Ave SW
Seattle, WA 98136
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rabbitmq-discuss