pbchase
commented
1 year ago Backup method
I've gone back and forth on the type of incremental backup we should make. The choices are to read the log data via the API (see https://github.com/OuhscBbmc/REDCapR/issues/320), read the log data directly, or read the updated records via redacp_read. After much vacillation, I think the best approach is redcap_read, while reading record deletion data from redcap_log_read.
Pros
- Full backups and incremental backups of a project to be made 100% via the API if an API token is available.
- A full restore dataset could be reconstructed by binding date-stamped rows, sorting by date, distincting, and then removing any rows that pre-date their deletion.
- At point-in-time restore can be created by filtering the data fed full-restore process.
- Each backup file is in the appropriate format for REDCap data input and redcap_write.
- Each backup file is restorable without processing.
Provide for project level backup of redcap data