I have created https://issues.apache.org/jira/browse/METRON-843.
Right now, if the RPM phase of the build pipeline fails, maven just plows along. I don’t believe this is correct. If the RPMs fail, the build should fail ( if it was trying to build rpms ). So for example, if you do vagrant up in full_dev, but the rpm’s fail to build, you won’t fail until installing the enrichment service in ambari ( because the /localrepo is empty ). I am not sure how we address this, but I think it is important that the build doesn’t continue in a failed state. Also, At one point this workflow worked : https://cwiki.apache.org/confluence/display/METRON/Metron+Installation+on+an+Ambari-Managed+Cluster Using a docker image to build the product, and then being able to deploy etc. I don’t believe this flow will work now, because the build will try to run docker ( within docker ) to build the RPM’s. I am not sure if this is technically true, but I think this is a regression. But the ambari/rpm flow is new, so that is OK, if we have a plan on how we are going to move it forward and iron out the kinks. Both of these issues are important ( I think ) on their own. Taken together, it seems to me that we may want to step back and think about how how we are using docker and refactor some of these disparate pieces into a more consistent approach. Thoughts?
