Erasing A Disk

Home » CentOS » Erasing A Disk
CentOS 8 Comments

Folks

I’ve encountered situations where I want to reuse a hard-drive. I do not want to preserve anything on the drive, and I’m not concerned about ‘securely erasing’ old content. I just want to be able to define it as an Physical Volume (in a logical volume set), or make it a ZFS disk, or sometimes make it a simple EXT3, ExFAT or NTFS
disk. However, old ‘signatures’ get in the way and Linux sometimes refuses to let me proceed. I know that a fool-proof solution is to use the “dd if=/dev/zero bs2768 oflag=direct” on the disk, but when we’re talking USB-connected hard drives of 8 TB, that’s an operation that can take days.

The disk in question might even have been corrupted. This would make using ‘zpool destroy’ to clear out a ZFS disk, or

I’ve tried erasing the first megabyte of the disk, but there are ZFS
or LVM structures that get in the way. So, does anyone have an efficient way to erase structures from a disk such that it can be reused?

Something like
-erase first N blocks (block defined as 4096)
– Erase blocks starting at block
– erase last blocks

At least such an algorithm would be quicker than erasing 8 TB of data.

David

8 thoughts on - Erasing A Disk

  • GPT for sure has backup metadata on the drive, so you won’t be wiping it by trying to remove the first few MB. You always should try using tooling made for the purpose instead of trying to manually do it. In this case try a tool like wipefs.

  • At Mon, 14 Sep 2020 13:14:44 -0700 CentOS mailing list wrote:

    Use dd in a script:

    #!/bin/bash
    # erase N 4K blocks starting at M
    # (M=0 means from the start of the disk)
    # usage: $0 start4Kblock numberof4Kblocks drive M = $1
    N = $2
    rawdisk = $3
    dd if=/dev/zero bs@96 oflag=direct count=$N seek=$M of=$rawdisk

  • If it is a Seagate, don’t bother. They have the highest failure rate in the industry.

    Look at the SMART statistics before deciding to re-use a disk, especially “Reallocated_Sector_Ct,”
    “Power_On_Hours,” and run the “extended SMART test.”

    Todd Merriman Software Toolz, Inc.

  • At 02:36 PM 9/14/2020, you wrote:

    Todd

    The reason I’m reusing a disk is not because of hardware failures, but rather because I’ve abandoned a particular use for that disk, maybe changed operating systems, maybe tried something using ZFS and want to change to some other techonology. I know that the best way to reclaim a disk after read-failures is to wipe it with zeros, and this takes days.

    And by the way, I typically use only WD disks, not seagate.

    Thanks for the confirmation.

    David

  • I’ve never run into a system yet where using dd to write zeros on the first few megabytes didn’t completely wipe the disk as far as the OS and existing file systems are concerned..

    dd if=/dev/zero of=/dev/sde bse536 count24