Hi Kevin,
basically you can use the spring integration test aproach.
http://static.springsource.org/spring/docs/2.5.x/reference/testing.html#integration-testing
So you use spring to start the context.
Another aproach is to use the camel testing support:
http://camel.apache.org/testing.html
Then the next thing is to feed the data into the route.
If you want to keep your file endpoint you could copy files into the
directory it listens to.
I normally use a producer (see
http://camel.apache.org/pojo-producing.html). For this aproach you need
to start the route with a direct endpoint.
In any case I would use the mock endpoint to wait for the number of
exchanges you expect to come. Additionally you can set a time limit. So
you can check the time part of your performance.
I have no idea though how to test for memory consumption.
In any case it is a good idea to have a look into the tests of the camel
components they show nicely how to test camel routes.
Greetings
Christian
Am 06.02.2010 17:09, schrieb Kevin Jackson:
Hi,
I have a route configured in Spring and I would like to integration
test (from file endpoint with production data to persistence in test
db)
For example:
<camel:route id="integration-test">
<camel:from ref="file-consumer"/>
<camel:split streaming="true">
<camel:tokenize token="\n"/>
<camel:bean ref="Unmarshaller" method="unmarshall"/>
<camel:bean ref="Transformer" method="transform"/>
<camel:to uri="mock:result"/>
</camel:split>
</camel:route>
Now I want to run this route with production data and ensure that the
sample of data I used in my unit tests is representative and that the
production data doesn't cause my pojos to break in any weird ways.
I'm also interested in seeing how much ram this single route will
consume with realistic data before I add it to the live routes.
# How do I trigger this route to run (in the context of an @Test
decorated method)?
# How do I make any meaningful assertions against the mock:result?
Essentially I want to 'assertNoExceptions' and assertTrue(memoryUsed<
80Mb) (or any other arbitrary value. If I wire up the db layer I can
assert against the data in the tables, but that's a data centric
approach and I think what I'm after is not easily defined by data.
For example I can prove by induction that if the code works for 1
valid input dataset then it will work for n valid input datasets, so
I'm leaning more towards testing intangibles (performance
[time|space], robustness of handling real workloads/data etc).
Sorry I can't really explain why I feel I need to do this, just that I
think it's necessary in this one particular case - does anyone have
any suggestions for how to at least start up the route in a testing
context?
Thanks,
Kev
--
Christian Schneider
---
http://www.liquid-reality.de
