Dirvish Filespace
Dirvish normally stores its archives as multiply-linked files in dated directories, as explained elsewhere. You can examine the contents of archives using filesystem commands such as 'du' and 'ls -liR'. This is useful to find out which snapshots are taking up the disk space. For instance, doing 'du -sh *' in a directory containing the snapshots will print raw disk space used by each snapshot. It will count each multiply-linked file only once, of course, so the actual snapshot footprint will be correct. For example, the output might be:
31.7 GB 20090705 0.6 GB 20090805 1.2 GB 20090905 19.0 GB 20090910 1.5 GB 20090915 1.9 GB 20090920
which could have resulted from backing up a 32GB directory on July 4, doing monthly incremental backups for a while until some big changes around September 10, followed by more frequent incremental backups in September.
Much of this information can be gleaned from divish log files, but it's useful to be able to check it independently.
The files themselves can be examined by 'ls -liR'. The [http://www.research-service.com/custom-term-paper.html term papers] files that weren't modified will be linked in from previous snapshots, and will have a higher link count (third column of the listing):
27093038 -rw-r--r-- 7 root root 15064 2008-01-03 15:34 data.dat
Here, the link count is 7 so this particular file is shared by seven snapshots.
The modified files will be saved on September 10 in directory 20090910 and will have a smaller link count, starting at 1 on Sep 10 and possibly increasing as time goes on, and the file is being linked to later snapshots.
To positively determine which files are linked together, you could list it in all the snapshots, 'ls -li */data.dat' . The shared files will have the same inode number (first column in the output), and a link count corresponding to the nuber of files with the same inode number.
