scripts/s3_media_upload ignores media files that have been deleted, but then pulled back into the local media store

[squahtx]

When (re)generating thumbnails, Synapse will ensure that the original image is in its local media store by downloading it back from storage providers. The s3_media_upload script does not notice when this happens and won't re-delete such files.

[H-Shay]

For my own edification, what would be the impact of this? Storage space being wasted?

[squahtx]

That's pretty much it. The script is supposed to offload media to S3 to save on storage locally.

[H-Shay]

Thanks for answering!

[kpfleming]

It seems like a workaround for this would be to delete the SQlite3 database that s3_media_upload uses as a database cache; it will no longer know what it has previously deleted, and will have to regenerate that list. It will take longer to run and generate more activity for the Synapse database.