Re: Extremely large level 1 backups with dump
On 20101201_215849, Peter Tenenbaum wrote:
> I've been using dump to perform backups of my home Debian workstation (I run
> squeeze, btw). I do a weekly level 0 dump and daily level 1 dumps.
> For some reason the level 1 backups are almost as large as the level 0 (the
> level 0 is 57.9 GB and the level 1 is 51.6 GB), even though we clearly don't
> modify anything like that many files in the interval from one day to the
> next. The commands that I am using for these backups are of the form:
> dump -0u -f "/media/FreeAgent Drive/filename0" -A "media/FreeAgent
> Drive/archive0" /dev/md3
> dump -1u -f "/media/FreeAgent Drive/filename1" -A "media/FreeAgent
> Drive/archive1" /dev/md3
> where /dev/md3 is my RAID1 array root partition.
> Any idea why the level-1 backups are almost as large as the level-0?
> Thanks in advance,
I don't use dump. I use rsync. Look into switching to rsync. With it
I do a full backup of a 90 GB source tree every day. It takes under
three (3) minutes each day to produce the new image of the full tree.
When looking at rsync, notice particularly the --link-dest= option.
This can be used to designate the backup from yesterday as a seed for
the backup being done today. The first time it is run, the time to
completion is similar to that of level 0 dump. But subsequent runs
are much faster. You can do several level 1 rsyncs per day, if you
wish. Use the -v option to tell rsync to tell you what it is doing.
It will show a list of only those files that actually needed to be
transfered, and a calculation of how much 'speed up' was achieved.
Most of the new image is hard links to the same file in an older
image. Since my data is pretty static, I have 300 daily backups in
disk space that is about 110 GB. I use a 500 GB USB external drive
for backup. Long before I run out of space, I will retire the disk
merely because it will have become too old to be kept in use.
Paul E Condon