[[ solved ]]
I have a stack of SATA hard drives that I need to erase.
I have a USB drive dock, a box that a drive can be set into that connects to my computer via USB-3.
I am using DD to write zeros to the raw device, in this case, /dev/sdf.
No matter the actual size of the drive dd stops at about 3 to 7 gb. These are 300 gb to 3 TB drives.
I am not mounting the drives, but I do ensure they are visible to the system with lsblk. To change drives I turn off the dock. The drive then disappears from lsblk. When I insert a different drive and turn the dock back on again /dev/sdf re-appears.
Are all my drives bad? If they are I will need to have them “professionally” destroyed at about $25 a drive.
Next Update –
I started with a USB to SATA adapter that looked like a small box with a SATA connector on one edge and a USB cable coming out of one side, it had a power supply that connected to the small box - everything out in the open.
Then I went to a drive toaster - a dock where you slot the drive into a hole in the top of the dock, again powered and USB-3 (blue connector)
As of this update I have opened my USB-3 external drive and removed it’s native drive and put in one of the 1TB drives I wish to erase. I also switched to my production laptop. Now I have issued a dd command and it has written so far 28GB from /dev/urandom.
I think this will finally work. - I am marking this solved.
I am a private person without any state or corporate secrets in hand. I do not do online banking.
My threat model I believe is limited to random drive-by actors.
I was hoping to be able to provide these drives to others to use, the screwdriver and hammer will render them into E-waste. But on that issue, once the platter assembly is disassembled and the platers are separated and mixed up the data on them is probably not recoverable? With the given that each drive has multiple platters.
A few gigs of zeroes will prevent random drive-bys. At that point the partition and filesystem table of at least the first partition is overwritten and you “can” recover files off it but you’ll be missing filenames and at least half the files will be corrupt due to fragmentation losing track of which files are where.
I agree with Ono that shred is a good tool for this. If you don’t want to use that, try increasing the block size to at least 1M if not 16M to reduce the overhead.