Reuse files across jobs
Attaching one file to twenty jobs stores it once. This page covers how that affects storage charges and what happens when you delete a shared file.
Attach an existing file to a new job
During job setup, choose Add from cloud storage instead of uploading. You can pick anything already in your My Library or a previous job's folder.
The file is referenced, not copied. Your storage usage doesn't change.
Promote a job's output to an input
To chain runs: one job's results feeding the next:
- Open the finished job and go to Results.
- Select the files.
- Actions → Make selected files available as input files.
They become available in your file picker for any new job. See download your results for the full workflow, including cloning.
Cloning a job doesn't duplicate its files
Clone a job ten times and its input files are still stored once. Only new uploads add storage.
This makes cloning the cheapest way to run a parameter sweep by hand, though for anything systematic, use a DOE or multi-run job instead.
What this means when you delete
Because files are shared by reference, deleting is less destructive than it looks: with one important exception.
Archiving is not the same as deleting a job. It sends files to the trash, and if the job had CFS mounted it permanently deletes the CFS files with no recovery. See archive and delete jobs.
Attaching very large numbers of files
Attaching more than about a thousand files to a single job needs a platform setting enabled for your workspace. Contact support@rescale.com if a large attachment fails.
For datasets this size, consider whether CFS fits better, mounting a filesystem avoids per-file attachment entirely.
Related
- Browse your files — where reusable files live
- The file lifecycle — what deleting actually does
- Download your results — promoting outputs to inputs
- Cloud Filesystem (CFS) — the alternative for large shared datasets