niclaslindstedt / docker-backup Goto Github PK

Why not?

Every backup session should get its own error log, that can be used to output the error log to e.g. a notification.

This is to prevent simultaneous backups/restorations/lts copy operations that cause conflicts.

For some reason, it's not possible to enter the passphrase for gnupg to decode an archive manually.

There is code now that allows you to "try yourself" if you e.g. try to restore a backup that was encrypted with a different password. But it fails for some reason.

I can restore the backups manually with the correct passphrase, but it's not working with the script.

https://github.com/niclaslindstedt/docker-backup/blob/master/lib/common/encryption.sh

We should verify the settings provided on startup.

Some settings depend on other settings, and should exit the program if anything seems invalid.

There should be an option for this since it will probably take a long time on large backups.

The only way to verify is to actually decrypt and then double-check the sfv for the original backup.

VERIFY_ENCRYPTION true/false

Changelogs (and releases) should be automated.

Just as we backup stuff to /tmp before moving them to /backup, we should do the same with pre-backups.

Instead of doing hashes, it would be good to use gnupg to produce signatures.

Of course, using sfv checksums should still be possible for systems where attacks are very unlikely, since creating signatures requires that the user provides a gpg key.

It would be nice to make this container working with standard protocols like webdav and ftp.
Env's are also perfect for this situation.

As we're closing in on v1.0, we should start using GitHub actions instead of relying on Docker Hub's build function. That way we can have version control here at GitHub instead of just using the :latest tag.

It would be nice to have notifications when the backup is completed, or have some errors.

Add a pre-commit hook that enforces conventional commits.

Also add a GitHub workflow that enforces PR titles to adhere to conventional commits.

With Slack and Discord down. Only e-mail remains :)

Currently you can only restore backups from the /backup volume.

If the backup doesn't exist there, we should use the /lts volume.

Instead of just writing logs to the output log, create a structured log system on the filesystem to make it easier to do queries for data and to generate reports.