I had a server with faulty drive in a Raid 6 array. Pulled out and back in the disk did not start the rebuild.

The same disk was swapped with another one in a backup server (same configuration as the first server) and the rebuild started on the backup server..

The disk removed from the backup server is now installed in the first server and both server's Raid have rebuilded successfully and works till then (a month).

The question is, why and how did the first Server Raid's controller recognized the bad disk while the backup server accepted it with no problem?

Is the information about a bad disk is written somewhere on the controller side?

Is the faulty disk really defective? Or some "software" error trigged the error?


  • 1
    Please edit the question with these answers : What controller on the first server, what controller on the second? What firmware on each? In general, when a hard drive fails in the slightest, I suggest never use it in production or production backup anymore because its likelihood to fail again soon is high. You can still use it in an unimportant computer, or recycle it. – Christopher Hostage May 22 '18 at 15:55
  • 1
    "Pulled out and back in the disk did not start the rebuild." This is bad practice. Rely on your alerting systems to tell you when something is wrong, and don't put the disk with the problem back in. – Christopher Hostage May 22 '18 at 15:56
  • @ChristopherHostage Thanks for your feedback. Both server have PERC H710P controller. I dont have the informations about the firmware version installed but I'm pretty sure they use the same. – Frédéric Harvey May 22 '18 at 16:51

Over time, the magnetic media of the hard drive's discs wears down. This means it is less responsive to induced changes in magnetism, meaning that eventually the drive can no longer use the media to store information. The rate at which the the media wears depends on many factors, and since some areas of the disc are used more than others, not all areas of the disc will wear at the same rate. This means that eventually there are some "Bad Sectors" which are no longer useful. Usually the data is moved to a good sector before it is unrecoverable, and the Bad Sector is no longer used.

The drive maintains a list of Bad Sectors in its SMART logs (specifically, the Grown Defect List aka GLIST) and reports these values to the RAID controller's firmware. The controller firmware likely has an arbitrary threshold of Bad Sectors required for a drive to be deemed "Bad." So it is likely that the drive passed the threshold in one machine but not in the other. Unfortunately, I can't give more specific feedback without knowing more about the controllers.

However, once Bad Sectors start appearing, they will appear more and more frequently as the media degrades further. Because of this, it is recommended to replace a drive as soon as you see a single Bad Sector appear.

Edit: typos

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.