Gratia Probes

We plan on installing probes on the following:

1. Prairiefire (Linux cluster, non-grid).  Install done; successful.
   
2. Merritt (SGI Altix, non-grid).  Install done; successful.
3. red (Linux cluster, grid-enabled).  Install done; successful.
4. dCache.  Install done; successful.
5. Hadoop (distributed file system; limiting probes to Globus GridFTP transfers for now).  Install done; successful
   

Gratia Collector

1. Ability to summarize usage by user and/or group.  This is going to be implemented by mapping the user to a VOi and the group to voc; this should allow summarization on both, right?  Not completed.
   
2. Interesting metrics (based on metrics we have found interesting :
1. Number of CPUs per job.  Not recorded in the summary tables. (FAIL).
1. We're told this will be fixed in the next Gratia collector update.
   
2. Job walltime / CPU time.  (PARTIAL - we want the walltime * (# of cores), gratia just records walltime).
1. We're told that this will be fixed in the next Gratia collector update.
   
3. # of GB transferred.  (PASS)
   
4. Job quality / wall success rate (PASS)
5. Transfer quality. (PASS)
6. Commonly accessed file names (FAIL; not even recorded in DB at all)
3. Replicate to OSG collectors based on probe name (PASS)