Distribution Computing SW

[jwebb <jwebb AT bnl.gov>]

Hi All,

We will skip today's meeting and resume regular meetings next week.

A quick summary of work over the last two weeks.  Thanks to Tadashi,
Paul and Xin, PanDA has been modified to better support the sPHENIX
workflow.  Specifically, a regular expression syntax has been 
implemented
by which users can specify the files which are to be saved in an output
dataset.  When this syntax is used, the files are saved with the 
matching
filename.  This differs from the 'wildcard' matching syntax, where 
matching
files are added to a tarball which PanDA ensures is unique.

We have tested this on the BNL test instance with both simple "hello 
world"
type jobs, and with realistic sPHENIX workflows.  At present, the 
calorimeter
and global reconstructions workflows have been run at modest scales (100 
jobs
at 400 events / job).  And the tracker workflow is running, but with 
filename
collisions which need to be understood / fixed.

Xin is working on deploying the updated panda server into the producion 
instance
today.

Cheers,
Jason