Then Guenther installed the federated database, by copying the FD prepared at CERN and doing an ooinstallfd.

Some work went into the setup of a script for the automatic job submission. The ideas for the jobs and database structure were:

      --> 1 edir(=epio) file should correspond to one database file.
 
      --> for the MC all events should be written, for data only CLAS 15,16,20,24

      --> the directory/file structure for the databases is
 
              $DBDIR : contains the bootfile and the FD file
              $DBDIR/mc/92/y12345_2.ALEPHDB.DB
                                  .....
                        93/
                        93/ ...
                     data/92/.....
                          93/.....  etc.
       

Finally the first jobs were submitted on Friday evening.

Problems encountered were:

Batch System : first wanted to use the PC cluster and the batch system CONDOR, but this caused some problems

Then decided to use a new 2-processor machine and do everything locally there, since also the disks are there. However, the machine was only delivered during the week and installed..

an inconsistency between the EDIR and EPIO files found in Munich (for 92)

problems related to the differences RH5.1 and RH6.0

Finally Guenther has given some first performance numbers. The writing rate of 150 Mbyte/hour/job is rather low. Andreas and Dirk gave suggestions for improving this considerably. Generally we will need about two weeks for writing everything.

Arno will do the babysitting of the production. When time permits, work should go into the read application, but only after the LHC++ people have solved the Linux problems.