Every now and then we see intermittent test failures, and rather than sweeping them under the rug, we should have a simple method to track and handle them. I started creating Jiras for tests that I've seen fail, but that don't fail consistently, or even fail more than once. For example, https://issues.apache.org/jira/browse/METRON-1851.
I think we're all taking steps to varying degrees already, but I want to call it out formally. I propose we create a ticket and add the label "test-failure." It might also make sense to send a quick note to the dev list or Slack channel, so attention can be brought to it and anyone else that may have run into an issue with the test can chime in. We can clean them out every few months - maybe do a review going into a release and close any that have not been reproduced for some time. What do you all think? Mike