2

We have a Linux server running with Software RAID1. We see some issues in /var/log/messages like: unreadable sector. I want to perform a complete fsck on the drive to get some more information, but a fsck /dev/md0 brings a clean due to the Software RAID layer in between. How can I check the real hard drive? Do I need to disassemble the whole RAID? How do I deal with the inconsistency in the partition due to the additional Software RAID header?

Does anyone have a good idea for this?

Khaled
  • 35,688
  • 8
  • 69
  • 98
Rick-Rainer Ludwig
  • 123
  • 1
  • 8

1 Answers1

2

fsck tests the filesystem, but this should be fine as long as the RAID hasn't failed completely.

To test the disk itself, use badblocks (in package e2fsprogs).

badblocks /dev/sda
Sven
  • 97,248
  • 13
  • 177
  • 225
  • It's ok to use `badblocks` without `-n` or `-w` on online filesystem, since it doesnt' do any writing. Changing any data on disk doesn't affect badblocks behaviour. – DukeLion Sep 19 '12 at 09:05
  • Yes, you are right, I misinterpreted the `-n` option. – Sven Sep 19 '12 at 09:08