Document (or remove?) requirement to run tests in own VM if multiple test classes using LightblueExternalResource

[alechenninger]

Taken from discussion in #231

You have to tell surefire to run the tests in a fork for this to work:

  <plugin>
    <groupId>org.apache.maven.plugins</groupId>
    <artifactId>maven-surefire-plugin</artifactId>
    <version>2.19</version>
    <configuration>
        <!-- forkCount is 1 by default. -->
        <!-- <forkCount>1</forkCount> -->
        <reuseForks>false</reuseForks>
    </configuration>
  </plugin>

This should probably be documented somewhere. Another solution would be to find a way to remove this requirement.

Thanks!

[dcrissman]

There are multiple examples of tests running sequentially. Are you running tests simultaneously? Could you provide an example?

[alechenninger]

multiple test classes

[alechenninger]

Sorry, the example is described in #231. If you have multiple test classes using the resource, the second test class to run using the resource will have problems if the tests are not run in a fork.

[alechenninger]

I understand this a little better now. The forkCount is 1 by default. It's the reuseForks bit that is actually important. It makes each test class run in a new VM, which means each test class individually triggers VM shutdown hooks, which causes the appropriate shutdown steps to happen before a new mongo/lightblue-rest can be stood up.

A fix would be to make those not VM shutdown hooks and just call them at the end of the rule like normal rules.

[alechenninger]

Another fix would be for the rule to detect instances of mongo and/or lightblue-rest are already running and reuse them.