Currently, when the backup tool is used to backup multiple backends, the backups are given the same ID, the user can provide a custom ID.
This grouping mechanism may be a useful feature but we do not see it as a core backup functionality, especially because the backups are not effectively taken at the same time. Backups are taken one after the other. when backing up large amounts of data or when backing up over the network, there can be a substantial time gap between each backup.
For now, backup files and code architecture rely heavily on backup IDs:
- metadata file names include the backup ID
- There are registry files just for listing all the backups associated with a backup ID
The goal of this issue is to keep offering the possibility to tag multiple backups without needing the whole code architecture and backup directory files to depend on it.
Stop using backup registries files.
Make --backupId truly optional: if none is provided, no ID should be generated.
File names should contain a timestamp or UUID in place of the backup ID. The backup ID is already stored in metadata files.
For now, it is not possible to use the same backup ID in multiple backup commands. Once this issue is resolved, backup tool should only verify that, given a backend, there won't be two backups with the same ID. In other words, the following should be possible:
Then running this command again should fail: