Re: [rdiff-backup-users] rdiff-backup when disk is full (was: Re: Trace when listing backup and not enough permissions)

[Frederik]

On 5/31/07, Charles Duffy <address@hidden> wrote:
> Frederik wrote:
> > Actually, rdiff-backup is a disaster in dealing with full disks:
> > looking at the size of rdiff-backup-data, a lot of space is used by
> > old increments. Unfortunately, rdiff-backup --remove-older-than fails
> > because no more disk space is available! So this leaves me with a
> > chicken and egg problem: I have to make free space, but to free space,
> > I have to remove old increments, which is impossible because the disk
> > is full :-(
>
> This is a good reason to have rdiff-backup run as a non-root user under
> normal conditions (such that when doing administrative commands as root,
> your root-only buffer on the drive [you've left that turned on in the
> filesystem options, right?] is available to allow the working space needed).

AFAIK only ext2 and ext3 reserve 5% for root, other file systems don't.
And no, this system does not use LVM, as this machine was installed
some time ago when LVM was still not as well supported as now by
various distro installers (and this machine was not even installed by
me).

Anyway, I deleted some old backups in rdiff-backup-data by hand. I'm
curious how it will backupninja will handle this now.

Still a good solution should be worked out for this kind of problems.
It is far from exceptional that disks become full when they are used
for back-ups, and it gives me a very unpleasant feeling that I cannot
trust on my back-up system to gracefully handle this situation
(without backtraces and without forcing me to delete random backup
files without using rdiff-backup itself).

Why does rdiff-backup need disk space to remove old increments?

--
Frederik