Currently we have a "retry" step in the Jenkinsfile when running against AWS Elasticsearch, which has a tendency to break connections and make tests fail unexpectedly.
It works, but re-executes all tests whenever any of them fails. This means we can waste dozens of minutes for each run. Sometimes we waste enough time to go over the 1h timeout...
Surefire/Failsafe 3.0.0.M4 introduced a parameter to automatically re-run flaky test, and apparently it only re-runs the few tests that failed. Let's try that?
We would only enable this option when testing against AWS Elasticsearch, of course.