The Design And Implementation Of Automation Test System For VMware Serengeti

VMware Serengeti mainly deploys and manages the Hadoop cluster in vSphere. With the increasing complexity of Serengeti’s functionality, traditional manual testing takes a long time, and configuring the cluster properties manually is too complicated. Therefore, by building automation test system, we can batch run test cases to complete regression testing, also complete test steps automatically. By this way, we can improve the test efficiency, and ensure product quality and stability.The automation test system mainly simulates Serengeti user’s behavior. Firstly, configure the specification file, then create the Hadoop cluster, and verify the attributes of each node finally. The system uses TestNG as test framework, so we can complete parametric test, method dependency and exception handling by annotation. The system is divided into3layers, including framework layer, component layer and test cases layer. Basic framework layer mainly encapsulates the basic test interface. Component layer includes entities, utilities and verification components. Test cases layer are functional test suites. In verification process, some cases related to node’s properties will call vSphere API, and some cases related to Web UI will use Selenium’s XPath to locate the Web page’s keywords. We also use Jenkins, a continuous integration tool, for smoke testing and regression testing, to ensure the stability of automated test cases.The automation test system has been applied in scrum. Automation rate of test cases is increasing effectively. For example, Sprint1’s automation rate has reached100%, so we can shorten the time of manual testing greatly. Smoke testing mainly ensures the major functions of Serengeti can work well when product code updates. By regression testing in daily build, we can find the defects as soon as possible, to avoid problems in product integration.