Project

General

Profile

JLab Farm Information » History » Revision 9

Revision 8 (Stephen Kay, 01/22/2021 02:56 PM) → Revision 9/10 (Stephen Kay, 01/25/2021 01:25 PM)

h1. JLab Farm Information 

 Included below is some information and resources regarding working with the PionLT data on the JLab farm. 

 h2. General Farm Resources 

 * "JLab SciComp Info Pages":https://scicomp.jlab.org/docs/ 
 ** "Tape Storage System Info":https://scicomp.jlab.org/docs/storage 
 ** "Auger batch queueing system":https://scicomp.jlab.org/docs/batch 
 *** Note, auger will be replaced by Slurm eventually, we're still using auger for now 
 * "Check Job/Farm Status":https://scicomp.jlab.org/scicomp/index.html#/ 

 h2. Git Instructions/Info 

 * "Git Instructions":https://redmine.jlab.org/attachments/download/1173/Git_Instructions.pdf 

 h2. Summer 2019 Run Info 

 The Summer 2019 data was stored alongside the earlier KaonLT data. The    data can be found in - 

 * /mss/hallc/spring17/raw/ - Raw data for 2019 PionLT run 
 * /mss/hallc/kaonlt/ - Analysed data 
 * /volatile/hallc/c-kaonlt/ - Temporary analysed data 

 Access to these directories is via the "c-comm2017" group 

 h2. 2021 Run Info 

 For the 2021 PionLT run, a new group was created, along with numerous new directories - 

 * New unix group: 'c-pionlt' 
 * New Farm/Auger project: c-pionlt 
 ** Use this Project name for simulation (and later analysis) jobs on the Farm 
 * Disk resources: 
 ** /volatile/c-pionlt/ - directory, 6 TB high quota;    2 TB Guarantee 
 ** /group/c-pionlt/ - directory (100GB quota) 
 *** snapshotted hourly; backed up automatically, use for analysis scripts, etc. 
 ** /work/hallc/c-pionlt/ - directory (1TB quota) 
 *** NOT backed up in any way 
 *** Has some small-file/performance benefits (but this doesn't really apply to hcana/podd analysis) 
 *** Not super useful in a modern workflow (mostly historical), but it's there if you want it. 
 *** Not intended for rootfile output or to store extra copies of CODA files, etc. 
 *** Good for large source code bases (that are 'backed up' on an external git repo, for example) 
 *** OK for modestly sized simulation output perhaps, though /volatile is still best. 
 ** New Tape/disk allocations: 
 *** /mss/hallc/c-pionlt/raw - tape volume for raw data output 
 *** /mss/hallc/c-pionlt/analysis - tape volume for analysis output 
 ** CACHE INFO COMING SOON