Stress and Stability Testing

[jsquire]

Summary

In order to ensure that the Event Hubs client libraries are reliable and performant for real-world use, a stress harness was created with the ability to perform local runs and monitor the library. With a large refactoring now released, the stress harness should be revisited and adjusted to ensure that it is consuming the Event Hubs client consistent with best practices and, once adjustments are made, several multi-day stress runs should be performed.

Scope of Work

• Revisit the patterns used within the stress harness to ensure they are consistent with the current library API surface and the recommended usage scenarios. Any gaps or adjustments identified should be addressed.

• Perform as many runs as possible using a 72 hour target; during at least one of these runs, there should be efforts made to introduce transient issues around network connectivity as a means of ensuring the client library recovers as expected.

• The correctness and resilience of the client library should be monitored and validated; any defects uncovered should be addressed within the time box allotted for this work or a new issue should be created to track it.

Success Criteria

• The stress harness is an accurate representation of real-world usage scenarios for the Event Hubs client library.

• At least two stress runs of 72 hours or more have been performed, with the client correctness and resiliency validated.

• Any defects uncovered by stress testing have been corrected and have test coverage, time permitting. For those unable to be addressed in the time box allotted, a new issue should be created and its priority determined and recorded.

• The tests necessary for to validate any changes made have been created or existing tests have been adjusted and pass reliably.

• The existing suite of tests continues to produce deterministic results and pass reliably.