If you do a backup and it fails for some reason, like out of space or restart of host, then the backup never finish and the partial backup file stay in the backup directory without being listed under backups in HAOS.
Since backups are often several GB large, then this will quickly eat up harddisk space and even make the issue occur more often.
Since the partial file is not listed, then it means there is an internal list somewhere in HA that is updated.
I suggest that when a backup is started then the backup is added to the list with a tag, like in progress, unfinished or similar, and when it is done, the the tag is changes to finished.
That way it should be possible to delete failed backups or even download them and extract what files it still succeeded to add.
I know that it is possible to install the ssh addon and start that and navigate to the backup directory in order to delete the file.
It is an annoying work around and does not make it better that you have to do it, not just to delete a file, but also to actually check if a file is even there.
The Supervisor builds the list of backups by reading the metadata from the backup files. If it does not appear in the list, it probably means the backup is corrupted or similar. Did the Supervisor log has an entry about that file?
Backups are written without fsyncing all the time, this means if the power fails, things might be partially written or not written at all. But I agree, it would be nice if Supervisor doesn’t simply ignore them but informs the user about them (probably make sense to create a repair issue: e.g. like Detected incomplete or corrupted backup … with size …, do you want to delete this backup?)
I do not know if there were a supervisor log message.
I run an automated weekly samba backup and did not notice it on the day, so I had to search several days back in the logs.
The worst part of this is that the failed backups is stopped at the point where the disc is full and that means the full disc will cause all kind of other problems until i gets corrected.
Maybe just outright deleting it would be better, even though the partial backup can actually be used to restore the parts that was added to the tar file before the process crash.