LinuxQuestions.org - sfill - secure free disk and inode space wiper

Moderators, if this is unacceptable feel free to delete. I just typed it up for CentOs and it is easy for me to paste here as well. More than you wanted to know about sfill and bcwipe. I simply moved the sfill from Kubuntu to CentOs and it runs.

I. I took the file sfill from /usr/bin/sfill in my Kubuntu machine, which I got by sudo apt-get install secure-delete.

I moved it into my CentOs 5.3 machine, also at /usr/bin/sfill using a flash drive, and in su mode, cp to final source.

II. Then, I cloned my emachine, which has several distros on it, to an external HD. I did this with a Puppy Linux live CD, which I booted;

(Of course, I had the USB lead of external HD connected to USB port, and running)

When it booted, I hit F2, then typed: puppy acpi=off pfix=ram

This loads the Puppy distro solely in Ram so it touches nothing.

A few clicks for keyboard, and when it shows Desktop, if you want change resolution.

Then click on the icon which says console. Enter dd if=/dev/sda of=/dev/sdb

These are the names Puppy shows on the Desktop.

Wait 16+ hours. Record message on completion in case you must reinstall it from clone.

In my case, it said:

234441648 +0 records in
234441648 +0 records out

If you need to reinstall from clone, reverse the sdb and sda, and set blocks per man dd to that number recorded.

III. I re-booted into CentOs, which is on hda13.

Open a terminal, to su

type sfill -I -l -l -v /home

This wipes all free space in the partition on which /home is located and only does one pass, filling with random data.

Hit <enter>

It will say creating /home/oooooooo.ooo...

Open another console terminal, type df <enter> Note the usage.

Wait a couple minutes and hit df again, and you can see the partition filling.

On my partition, it was in 1K blocks:

11227664 7839068 used, 2809056 available 74%

It climbed toward 100%. When it said 100% it actually kept going, because the 100% is rounded up, I guess.

When it hit the end, it lets you know it is done. In a minute, the df will show 74% again.

On mine, it took 17 minutes and 14 seconds for one pass.

IV. This does not wipe slack space, the difference between cluster size and file size. Allegedly, this requires a program from http://www.jetico.com/download/ which is BCWipe-1.9-3.i386.rpm

I installed it with: rpm -i BCWipe-1.9-3.i386.rpm

I typed: bcwipe and it gave me the basic nomenclature. There is an option for wiping slack space.

Playing with it, I learned that this command will wipe all slack of unhidden files in the current directory only.

bcwipe -mz -n 60 -S -v *

In the Desktop it reported: Wiped 72KB of file slacks.

Note -n 60 is not needed for -mz since it is single pass.

Changing su console to, or example, /sbin, then run above command wiped 472K of file slacks.

It will not do recursive, only current directory.

Someone on the official forums said this in answer to a question how to clean all slack:

"You can try something like find <path_here> -type f -exec bcwipe -Smd {} \;
Remember, it will be lenghty operation if you have many files there."

I would not trust a command which said "something like" but I bet there is a way if that isn't correct.

So far no system damage noted. But, I sure feel better with a clone.

REMEMBER TO WIPE YOUR EXTERNAL HARD DISK WHEN YOU ARE DONE, IT HAS THE SAME FREE SPACE YOUR HD DID.