I am running the analysis in PROOF (currently in TProofLite). I have many input .root files (lets say file0001.root through file000n.root) each with same tree (Lets say myTree). Each tree has UserInfo() object (Lets say myObject) on which analysis depends. So I would like create a dataset in which, for each ‘file000i.root’ at some worker there should be an output file ‘outfile000i.root’ on same worker. How can I create such dataset in analysis.
TProofOutputFile creates only single outputfile per worker. Also if I understand right, TProofOutputFile assumes that data
is stored in a Mass storage pool which is different from worker nodes. In my case all machines would be serving as workers as well as data storage devices.
If I understand correctly, you would like:
- To force local processing, i.e. each worker processes only files resident on each
- Have one output for each file
The best is, probably, to have files has entries or similar, and have Process to get one file producing another. I will try to provide an example.
And to have a dataset created and registered out of all the locally created output files like TProofoutputfile.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.