[Feature Request]: Support backing up individual folders

[mirkohubtv]

Problem / Motivation

Some applications (e.g. Jellyfin) store data in plain files such as JSON, configuration files, or metadata inside specific directories. Currently, it's not always possible to include these folders directly in backups, which makes it harder to fully preserve or migrate setups.

Proposed Solution

Add support for backing up folders or files.

• Select one or more directories/files manually

Optionally, filters (include/exclude patterns) could be supported to fine-tune what gets backed up.

Alternatives Considered

Currently, the only workaround is to manually copy these folders outside of DBackup

Area

Database Adapters

Additional Context

This would be especially useful for self-hosted apps like Jellyfin and similar tools that store data in filesystem directories rather than databases.

[Skyfay]

Hey @mirkohubtv

Thanks for your feature request.

I totally understand the desire to include additional files in a backup, but DBackup is currently designed exclusively for databases. File backup would make everything much more complicated, especially if you were to combine them.

Currently, file backups should be handled differently.

Accordingly, I would not pursue this feature request at this time. Maybe it will be possible someday in the future.

I hope that's okay with you?

[mirkohubtv]

Hey, thanks for getting back to me!

I totally get where you're coming from

That said, a lot of popular self-hosted apps these days (like Jellyfin, Immich, etc.) don’t even use a real database. They just rely on folders full of JSON/other files. For those, a database-only backup doesn’t really help with a full restore.

Would you consider adding a simple, optional “File Folder” adapter? maybe even just an SFTP/SSH one? The idea would be pretty straightforward: user points it at a local path or remote folder, it zips/tars it up, and it gets backed up alongside the databases using the same retention rules.

It would stay completely separate from the database logic, so it wouldn’t mess with anything that already exists. I think this would make DBackup way more useful for the whole self-hosted community without turning it into a general-purpose backup tool.

Think about it. :)

[Skyfay]

Hey @mirkohubtv

Thanks for the thoughtful follow-up. I can see where you're coming from and I appreciate the suggestion!

That said, I want to be transparent about why even a "simple" folder adapter is harder than it looks, and why it doesn't quite fit DBackup's scope:

The consistency problem Unlike databases, there's no protocol to create a consistent snapshot of a folder while it's being written to. If Jellyfin is actively writing its JSON/metadata files during a backup, you risk ending up with a corrupt or partially-written archive and the user won't know until they try to restore. DBackup's DB adapters solve this by design (e.g. pg_dump, mysqldump guarantee a consistent point-in-time snapshot). There's no equivalent for raw files.

What "simple" actually involves a production-ready file backup adapter would need to handle:

• Locked/in-use files
• Symlinks and special file types
• Correct permissions and ownership on restore
• Incremental backups (zipping an entire folder every run doesn't scale)
• Deduplication
• A reliable restore flow (to where? with which permissions?)

These are all solved problems but in specialized tools like Restic, Borg, Duplicati, Veeam or ABB which are purpose-built for exactly this use case.

DBackup is intentionally scoped to database backup automation. That's where the value lies deep integration with DB engines, proper dump/restore flows, and consistent backups. Expanding into generic file backup would mean building a worse version of tools that already exist, while adding significant complexity and potential for silent data loss.

For self-hosted apps like Jellyfin, I'd recommend pairing DBackup with Restic or a similar tool they complement each other well.

I hope that clarifies the decision. Thanks again for the suggestion and for engaging with the project! 🙏

[Skyfay]

Hey @mirkohubtv

After giving this some more thought and conducting a poll on Discord, I've decided to start supporting file-based backups soon and, later on, Docker volume backups as well. The implementation is designed so that you can secure each destination adapter, if desired.

So you're completely free to decide whether and how you want to back up your data. You'll have the option to back up only databases, only file-based data, or both together in a single job. I think that's exactly what you wanted back then.

I've reopened this issue and will post updates here. I hope that's okay.

Have a nice day :)

[mirkohubtv]

you're the best!