Experiment data has been separated from user data as far as possible. New fileclasses to distinguish between primary accounts, secondary accounts and service accounts. do_not_migrate fileclass for data which should probably be deleted.
Summary
971 CASTOR users. Around 1/3 of them have <100 files.
total count of files: 61,765,238, total size: 4.9 PB
7 users with >100 TB. 246 users with >1 TB.
Only around 20 users who wrote data to CASTOR in the last year. These seem to be mainly experiment analysis or MC use cases.
Strategy
Clean up/delete unneeded files where possible
Migrate users with <1 TB to CERNBox. Currently if we total up files for all users with <1 TB it comes to 76 TB.
Users with >1TB, 2 options:
let them negotiate a higher limit
leave their data on tape and migrate it to CTA (last resort)
To Do
Need a policy for users who have left CERN and left files behind in CASTOR.