Running a Distributed Test

You run a distributed test as you would run any other test, by clicking the Run button in the SystemTest toolbar.

When you run a distributed test:

• Pre Test executes once, on the client machine (the machine from which you run the test).

• Main Test iterations execute on the cluster of worker machines defined by the user configuration.

• The SystemTest software waits for the distributed test to complete.

• If there are errors when the distributed test iterations run, only the first error from the tasks will be reported to the Run Status pane in the SystemTest software once all tasks have completed.

• At the end of each Main Test iteration, test results are saved and returned to the client machine once all Main Test iterations have finished executing.

Note   Because Main Test iterations run across a number of tasks, there is no guarantee as to the order the tasks (Main Test iterations) will execute. Tests should not be written with the assumption that iterations will execute in a fixed order. Also, because Pre Test runs on the client machine, and tasks run independent of each other, Main Test iterations should not rely on data persisting across multiple iterations.

• Post Test executes once on the client machine, after Main Test executes or has errored while running.

• Test execution reports are generated at the end of the test, if enabled by the test.

• Generated plots are not shown on the client machine while the test runs, but are captured and displayed in the Test Report. Note that plots generated on worker machines will only reflect information generated as part of the task. Plotting multiple data points or lines on a single plot will only reflect the data pertaining to iterations executed as part of a single task.

Note that MATLAB and the SystemTest software remain in a busy state until the distributed test is done running or is stopped.

Caution   It is recommended that you do not run a test containing hardware-related elements in distributed mode. That includes the Image Acquisition Toolbox element, the Data Acquisition Toolbox elements, and the Instrument Control Toolbox elements. These elements will likely error out because the connected hardware will not be available on the workers.

The following diagram illustrates the relationship between the SystemTest software, the scheduler, and the workers. Task 1 to Task X and Iteration 1 to Iteration N are determined by what is shown in the Distribution of main test runs across tasks section in the Distributed tab. For example, if the Distribution of main test runs across tasks for a test with 90 iterations is set to Default (3 iterations per task, 30 tasks), that means your test will execute 3 iterations for each of 30 tasks. In this case, Task 1 might run iterations 1, 2, and 3, and Task 2 might run iterations 4, 5, and 6, etc.

Distributing Iterations Across Tasks

Example: Distributing a Test

© 1984-2012- The MathWorks, Inc.    -   Site Help   -   Patents   -   Trademarks   -   Privacy Policy   -   Preventing Piracy   -   RSS