Improving the test framework

There are three areas of focus for improving Lustre's self-testing infrastructure. These three items can be complimentary and may (should) share some of the same features.

• Development of a completely new test framework that will provide the capabilities to test Lustre at exascale.
• Incremental benefits by improving current test framework.
• Providing better coverage by improving individual tests.

Currently, we are in the process of collecting the test framework requirements to facilitate the above. Background on the current problems and some possible solutions were presented by Chris Gearing and Nathan Rutman at LAD '12, available here. These were more fully developed in a Google doc here [Nathan can grant edit rights]).

We are also looking into alternatives for an upper-level (no Lustre knowledge) automation framework evaluation. This would be used to automatically deploy test resources and test requests, and track the results.