Difference between revisions of "CaFe Disk Space"
From HallCWiki
Jump to navigationJump to search| Line 46: | Line 46: | ||
;NOTES | ;NOTES | ||
* During online analysis, raw data from tape will be automatically copied over to cache. | * During online analysis, raw data from tape will be automatically copied over to cache. | ||
| − | * Cache has a certain quota, which if it is exceeded, the user will have to use the 'jget' command to copy over the files from tape to cache. | + | * Cache has a certain quota, which if it is exceeded, the raw data files will be automatically copied to tape. <br> |
| + | See cache data policy: https://scicomp.jlab.org/docs/filesystemsv2#cacheDataPolicy | ||
| + | * to recover the raw data back to cache, the user will have to use the 'jget' command to copy over the files from tape to cache. | ||
Revision as of 09:45, 1 April 2022
Disk space allocation directories for the CaFe experiment. For additional info on filesystems, see: https://scicomp.jlab.org/docs/filesystemsv2
Hall C cdaq-cluster filesystems
1. main directory (for online analyzer, DBs, replay scripts, etc.):
/home/cdaq/cafe-2022/
2. output directory (ROOTfiles, REPORT_OUTPUT, PDF files, etc.):
/net/cdaq/cdaql1data/cdaq/hallc-online-cafe2022/
- NOTES
- Please do take special care to avoid modifying any files outside of your ~/cafe-2022 directory.
- File storage directories during the online analysis of CaFe.
- There is a README in the first directory with a few notes.
- It is recommended to make a '~/cafe-2022/go_analysis_cafe' shell script that sets up your environment from scratch. For example, no changes/additions to anything in ~/bin, changes to init files like .tcshrc, .bashrc, etc, text editor config files, etc.
Farm/CUE disk allocations
/volatile/hallc/c-cafe-2022/ ( 4 TB high quota; 2 TB guarantee )
- NOTES
- Files are NOT backed-up
- Use for all large file output from analysis or simulation jobs. When guarantee threshold is exceeded it is possible to have files auto-cleaned up (removed). See policy: https://scicomp.jlab.org/docs/volatile_disk_pool
- Files you want to keep should be pushed to tape using jput
/work/hallc/c-cafe-2022/ directory ( 1TB quota )
- NOTES
- Files are NOT backed up.
- Good place for analysis software, database files, etc.
- These files should be in GitHub or have similar backups.
- Nothing that matters should be stored only on /work.
/group/c-cafe-2022/ directory ( 100GB quota )
- NOTES
- Backed up regularly.
- Best place for analysis/replay scripts; software that is being actively developed, etc. (But, still use GitHub!)
Tape allocations
/mss/hallc/c-cafe-2022/analysis
- tape volume for analysis output (ie. simulation output, replay output that you want to keep long-term, etc.)
/mss/hallc/c-cafe-2022/raw
- tape volume for raw data output (ie. production CODA files from the Hall)
- Raw data are NOT accessible directly from tape. To make them accessible, they must be copied over to cache.
Cache allocations
/cache/hallc/c-cafe-2022/analysis/
/cache/hallc/c-cafe-2022/raw/
- NOTES
- During online analysis, raw data from tape will be automatically copied over to cache.
- Cache has a certain quota, which if it is exceeded, the raw data files will be automatically copied to tape.
See cache data policy: https://scicomp.jlab.org/docs/filesystemsv2#cacheDataPolicy
- to recover the raw data back to cache, the user will have to use the 'jget' command to copy over the files from tape to cache.
