Hadoop integration testing

I would like to know what is the best way to perform integration tests in Hadoop ecosystem?

Currently, I use Hadoop, HBase and Oozie, and I was wondering what would be the best approach to test the integration. So I don't want a mock of Oozie or HBase, but I want a 'light-weight' instances of those so I could for example write to HBase from a web service, without the need to inject a mock. Similarly, I don't want a mock Oozie client, but light-weight Oozie running on some port.

Would it be a good approach to setup a pseudo-mode cluster on a single machine and install HBase and Oozie additionally, or is there a better way?