Re: [rdiff-backup-users] Unable to restore interrupted backup

[Dominic Raferd]

On 08/07/2011 13:30, abschiedsstein wrote:
> Hi,
> I used rdiff-backup for a while and everything was fine. Backups were fast and 
> restoring was no problem.
> Then, after an interrupted backup, the data was broken. Neither restoring nor 
> fixing with --check-destination-dir was possible.
> I guess there is somekind of a standard routine to fix this, but I was not able 
> to figure it out until now.
> Before I get into details, here are some information about my machine.
>
> OS: Ubuntu 10.10
> rdiff-backup: rdiff-backup 1.2.8
> The backup directory is on an external usb drive.(NTFS)
>
> When I tried to restore a file I got the message that the last backup was interrupted and that I 
> should run --check-destination-dir (so far, so good). I did this. There were many messeges like 
> "... metadata, but could not be constructed from existing increments because last increment 
> had type ..." . When I now run --check-destination-dir again, I get : "Fatal Error: 
> Destination dir /media/TREKSTOR/Sicherungen/neodata-backup/home does not need checking". BUT 
> restoring is still not possible:
>
> > rdiff-backup -r now /media/TREKSTOR/backupdir/Desktop/protokoll.pdf 
> > Desktop/proto.pdf
> > Warning: Could not restore file .!
> > A regular file was indicated by the metadata, but could not be
> > constructed from existing increments because last increment had type
> > dir.  Instead of the actual file's data, an empty length file will be
> > created.  This error is probably caused by data loss in the
> > rdiff-backup destination directory, or a bug in rdiff-backup
> I would like to know a way to fix the backup. It would be completely sufficient 
> if everything since the last successful backup would be deleted.
> I hope someone can help me. Thanks in advance.

I attach a bash script which automates the process of forcibly regressing an 
rdiff-backup repository. You can use this to override the 'does not need 
checking' message.

Help is built in, just run the script without parameters to see. It does
a single step regression, it can be run multiple times if you need to
regress the repository back further. All the usual caveats apply.

Dominic

rdiff-backup-regress.sh
Description: Text document