[Dirvish] dirvish not making hard links but copying files, consuming much space

Keith Lofstrom keithl at kl-ic.com
Fri Jun 9 15:52:46 UTC 2006

On Thu, Jun 08, 2006 at 07:10:59PM +0200, Bloombox Team wrote:
>  ...
> Dirvish inits a vault after running a dirvish --init --vault n41 and it also
> runs fine about 1 to 3 times after initialization by only storing about
> 100MB of files which have changed.
> After a few runs dirvish backups all files which is about 5GB each run - and
> this jus fills up my backup storage very fast ... :-( The problem is: Those
> files haven't changed at all from my point of view.

Dirvish is a wrapper around rsync, designed to minimize the complexity
of configuring rsync.  So the problem you are having is with rsync doing
something you do not want - copying all the files rather than the ones
that you think are changing.

The default behavior of dirvish and rsync is to decide a file has
changed if the size or the date has changed.  I would guess the
date has changed on your files.  Perhaps there is some program that
runs once a day that does a "touch" on the files.  You can check 
this by looking at the dates with "ls -l".  You can look at the
dates of the real files, and of the copies in the original and
the extra image.

You can tell whether the files really change by doing a "diff" an
individual file in the original image and in the extra image.  If
there is not enough room to store both, then you can experiment
with a very small vault with just a few files in it - do that for
a few days and see if you get the duplication behavior.

If the files are have a date change, and you cannot find out the
reason, you can run dirvish with the --checksum flag and see if
the files are still copied.  The --checksum is slower than looking
at the date change, but your vaults are only 5GB so that would not
take too long for you.  If the files are identical and the path
does not change, but they are still getting copied by rsync with
--checksum, then we have an interesting new problem and I hope
you will help us find that.

BTW, you can do us a favor.  Many of our new German friends are
here because of an article in c'T magazine.  Could you find a
URL for that article so we can add a link on the dirvish website
for it?  


Keith Lofstrom          keithl at keithl.com         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs

More information about the Dirvish mailing list