Marek Horst
Marek Horst
Hi @S6savahd, it seems you are not assgined to this particular project, you are among the OpenAIRE-core team members and the whole group is assgined. I've removed you from the...
Related redmine ticket: [#5291](https://issue.openaire.research-infrastructures.eu/issues/5291).
It will be nice to run some benchmarks to compare RDD-based solution with the dataframes-based one.
Once we introduce changes proposed by @johnfouf we could swap the joining order mentioned in https://github.com/openaire/iis/pull/1098#discussion_r445615428.
#1122 is one another related issue: in order to give user libs precedence over sharelibs/parcels in plain M/R jobs one should set this property in the `workflow.xml` definition: ``` oozie.launcher.mapreduce.user.classpath.first...
In spark, under current CDH 5.16.2 OCEAN cluster setup, it seems user libs have already precedence over sharelibs/parcels. As already mentioned in https://github.com/openaire/iis/issues/987#issuecomment-534184328 it can be also enforced using spark...
It turned out when the module is deactivated the (empty) rawset ID is not elected as the LATEST by the ActionManager. Therefore we can lower the priority for this issue.
Currently tested solution was based on `jBrowserDriver` project. Even though it works properly in development infrastructure it does not work (hangs) on IIS cluster probably due to the following issue:...
New pull request to be merged: #853.
Part of this task could be undertaken in the scope of #1154, at least in the context of bibrefs deduplication.