70 likes | 228 Views
Input for COCOA. Questions about the structure of the input system for COCOA James N. Bellinger 3-June-2008. Presuppositions. COCOA runs offline correct Cannot have two Oracle databases open I think each subsystem has a set of tables and different DB accounts (online at least)
E N D
Input for COCOA Questions about the structure of the input system for COCOA James N. Bellinger 3-June-2008
Presuppositions • COCOA runs offline • correct • Cannot have two Oracle databases open • I think each subsystem has a set of tables and different DB accounts (online at least) • COCOA output should go to one single DB • Want to keep COCOA input simple • correct
Data Sources • DCOPS data is in Online DB (direct) • Online DB to file writer being rewritten • Endcap analog is in Online DB (PVSS) • ? Not familiar with this • Link data is ? • OMDS Online DB (from PVSS) • Barrel data is ? • ? Not familiar with this This is a bit DCOPS-centric, because that’s what I’m familiar with.
Presumed Input Data Flow Composite Event File Calibration Database Event Synchronizer and DQM COCOA Barrel Data Nfs:/cms/mon/data ? Link Data Endcap Analog DCOPS Data Online To Offline Xfer Offline Online ? ? ? DCOPS Data /cms/mon/data ? DB to Root file Online Database
Things DCOPS chain needs • Data in DB DONE • Program DB File In progress • Scheduler DB File • Online area ? /cms/mon/data ? • Run on cmsmon and store root files in cms/mon/data/alignment • Can make 3 subdirectories for each subsystem (I hope) • Online Offline ? NFS+ copy? • Under discussion with B. Badget, F. Glege, P. Paolucci • Special option for our needs under discussion/development • Scheduler NFS COCOA area • Offline COCOA area • Aim is some disk accessible from the CAF • Responsible party for creation • Responsible party for monitoring • Divide responsibility by subsystems? Depends on nature of “creation” and “monitoring” Scheduler includes the “cron” scripts and auxiliary record keeping files
Questions • How does one write to /cms/mon/data? • You need to request an account on cmsusr0 and cmsmon • Is /cms/mon/data nfs-mounted across to the offline alignment node? (Slide suggested it) • YES, this was the solution B. Badget proposed. It is still not final. • NFS mounted on one single alignment node at CMS centre, offline, which we still need to define. My responsibility (Gervasio) • Does one have to copy data to another location offline? • Yes, if we want to run COCOA on the CAF • Can we run COCOA interactively rather than on CAF and still access the data files? • From our offline node our root files can easily be transferred anywhere • Who is writing the “alignment event builder?” • Nobody as far as I know • I was hoping COCOA could take as input 3 root files: one from each subsystem • If this is not the case we need to define this task better. • As we create components, who is the central person to accept the pieces and put them in the correct locations online and offline? • Each subsystem should assign a person to put their pieces in the correct locations, which we will share, but which can be divided into three subdirectories, for example
Alignment Data Flow CMSMON PC (CMS CR) S2G19 PC ROOT /cms/mon/data/ files.root PVSS RDB Kvaser online (P5) offline CMS Center Alignment Node Cron job NFS-mounted /cms/mon/data/ OMDS DB ORACLE CAF COCOA root input ?? Ask Pablo ORCON DTAlignmentRcd DTAlignmentErrorsRcd CSCAlignmentRcdCSC AlignmentErrorsRcd ORCOFF DTAlignmentRcd DTAlignmentErrorsRcd CSCAlignmentRcdCSC AlignmentErrorsRcd CMS RECO Gervasio’s slide