Hi all, We have been trying to find a sane way of running Oozie workflows that load other components' *-site.xml files, and it seems there is no standard way of doing this?
To give an example, we wanted to run a Sqoop action that uses the plugins defined in sqoop-site.xml. The most common solution I see is copying sqoop-site.xml to the sharelib (putting it somewhere else than the sqoop root, since it could conflict with the sqoop-site.xml file created by the Sqoop action), and then specifying its location in the <job-xml> element. A similar approach needs to be taken for Hive. One of the downsides is the configuration file needs to be updated in HDFS every time there's a change in the original one. The other approach is bundling the xml file in the workflow, but this needs to be done every time a workflow is deployed. Now, it seems to me loading a *-site.xml file for an action should be a common case. Would it make sense to have this logic in the action executors themselves, so that the configuration is always loaded during the action execution? Let me know if I am missing something, I'd be ok working on a solution if this makes sense. Regards, David Villegas.
