60 likes | 167 Views
C.Cheshkov 30/09/09 Weekly Offline Meeting. PROOF Reconstruction: input and output file handling. Raw-data input. AliRoot now handles the following syntax of input raw-data files collection: “collection://XXX.xml” - collection file generate for example by alien find comand
E N D
C.Cheshkov 30/09/09 Weekly Offline Meeting PROOF Reconstruction: input and output file handling
Raw-data input • AliRoot now handles the following syntax of input raw-data files collection: • “collection://XXX.xml” - collection file generate for example by alien find comand • “collection://XXX.root” - ROOT file with entry-list(s) generated for example by a tag analysis system • “collection://XXX” - text file generated by querying alien or any other means • In PROOF we usually use the last one: XXX file generated by a simple shell script, however other methods should also work
Raw-data input: proposal • Allow simpler usage (for the shifter): • “???:///alice/data/2009/LHCXX/run#/raw/” • “raw://run#” (similar to OCDB syntax) • “run#@alien://...” • Other proposals?
Output files • At the moment we have only AliESDs.root and Merged.Qa.Data.root as output files • The output location is controlled via AliReconstruction::SetOutput(“...”) • QA file follows the location of ESD file • The present interface is not flexible, doesn't allow storing of additional files
Output files • We can have quite a lot of various combinations: • Store merged ESD [and QA] on the client machine (when SetOutput not used), needs xroot daemon running locally • Store merged ESD [and QA] on alien in user's folderSetOutput(“alien:///alice/cern.ch/user/x/xxxx/AliESDs.root”) • Stored merged ESD [and QA] on CAF scratch space? • Store ESDs [+other files: QA,RecPoints,...] in a folder with one subfolder per worker (local,alien), not needed? • Store ESDs in a registered dataset on CAF • Store ESDs [+ other files: QA,RecPoints,...] in a registered dataset on CAF • Only the first two ways are implemented now
Output files: proposal • We need some simple and already established output files URI syntax (preferably an 'alien-like'): • “AliESDs.root,Merged.QA.Data.root” - on clien machine • “alien://..../AliESDs.root,alien://.../Merged.QA.Data.root” or “AliESDs.root,Merged.QA.Data.root@alien://....” - merged output files in alien • “AliESDs.root@dataset:MyRun” - ESDs in a registered dataset on CAF • “root_archive.zip:AliESDs.root,ITS.RecPoints,...@dataset:MyRun” - prepare root_archive.zip on the workers and registered the dataset containing all the requested files • Other proposals?