Troubleshoot clusters

An EMR cluster runs in a complex ecosystem that comprises open-source software, custom application code, and Amazon Web Services. When a problem occurs with any of these parts, the cluster might fail or take longer than you expect for it to complete. The following topics can help you identify cluster issues and how to fix them.

When you are developing a new Hadoop application, we recommend that you enable debugging and process a small but representative subset of your data to test the application. You may also want to run the application step-by-step to test each step separately. For more information, see Configure cluster logging and debugging and Step 5: Test the cluster step by step.