With builds failing pretty often, I think it's important that reproducible test failures be detected as such and alerted so that we can focus our build triage attention where it's most fruitful, and to reduce the time for the bug to be active.
There used to be a "repro" build that Steve Rowe originally worked on: https://ci-builds.apache.org/job/Lucene/job/Lucene-Solr-repro/ The corresponding script is: dev-tools/scripts/reproduceJenkinsFailures.py Clearly it hasn't survived the gradle transition. Does anyone have thoughts on reviving this vs some other ideas you may have? One idea I've had is to have the Jenkins job for the builds include a reproducibility step -- an encore for the failing tests to try to do their thing again. Then we could differentiate the build status this way -- Unstable vs Failure on the reproducibility. ~ David Smiley Apache Lucene/Solr Search Developer http://www.linkedin.com/in/davidwsmiley
