0

I have a IBM 3650X M3 with 2 disks in a raid1 array. One of them is in "failure predicted" state. I have two question about this case:

  • Should I trust it and change a working disk? How the system can predict a failure?
  • How to change a working disk? Is ther a procedure for this controller or I just need to remove the disk?
Tobia
  • 1,210
  • 8
  • 37
  • 73

1 Answers1

2

Failure prediction is generally based on utilization statistics, diagnostics and/or comparing bit data viability to its surrounding bits; in most cases, almost all hard drives have S.M.A.R.T. capability to let a system know if there are performance abnormalities on a hard drive.

For example; if the drive notices that the data in one sector is harder to read than others, it could indicate abnormal wear and tear. If the drive has stabilizers, and it notices the drive is wobbling, the drive takes that into account. If the drive is taking longer than normal to boot, to read, to cycle, etc...

In most RAID systems, so long as you have a hot spare, you should be able to just pull out the failing disk and replace it with a disk of the same make/model/size/speed. The system should rebuild the RAID based off parity bits automatically.

If you do not have a hot spare or hot swappable system in place, you'll need to go into the RAID management tools to and find the utility that allows you to manage your disks. From there, you'll need to eject the failing (or predicted to fail) disk through the utility. Some utilities will then require you to shut down the system to perform the replacement. After the disk has been replaced, you may need to manually add it to your RAID array and partition it as necessary, before using the RAID management tools to rebuild the array.

CIA
  • 1,606
  • 2
  • 13
  • 30