1

So I got this server and if has a visual alert on one of the drive.
After further investigating with storcli I notice that is has a few media error counts.

I did some investigating about this and apparently these sorts of error are uncorrectable damage on a disk sector so it is remapped in order to not use said sector.
But apparently almost all drives have these sort of errors ads they are also caused by manufacturing imperfections.

So I have several questions:

  1. If this is normal why is a visual alert shown just for a few errors?
  2. How can I view more detail about these errors using storcli
  3. Can someone explain to me what are the other error count and shield count

If I'm misunderstanding something please explain it to me.

Thank you

1

I confirm that media error counts means a physical sector gone bad, generally discovered during an application read or array scrub.

In this context, "a sector gone bad" means that the physical disk was incapable of reading the original sector, returning an error to the RAID controller. The fact that the HDD itself can mark the sector as "to-be-remapped" is transparent to the RAID controller, which will simply try to re-write the same sector by using data from the other mirror leg/parity. If this re-write fails (meaning no availability of spare sectors from the drive itself), the disk is generally marked as failed.

If sporadic, such read errors are not too alarming and in fact most RAID controller mark a disk as bad only after some error threshold is crossed. In other words, 1 media error will simply be reported, while 100+ errors will definitely also mark the disk as bad (or as "predicted to fail soon" state).

4
  • Just to confirm, no matter how new a drive is there will always be media errors? And when you say the HDD will mark the sector as to be remapped, does this mean that the HDD will redirect the write to the spare sector and not the RAID card? – Symbux Sep 17 '20 at 21:58
  • New drivers should have zero user-facing/SMART reallocated sector (the so called glist). They can have internal, hidden reallocated sectors tracked by the plist, but this is not user accessible nor reported by SMART (the plist never grow, is a factory written list). When a sector is marked as bad by the HDD, the RAID card will simply try to overwrite it and the HDD will remap the affected sectors with a spare one. If this operation fails the RAID cards marks the entire disk as failed, unless it has an internal bad block/remap list (recent versions of MDRAID have it, for example). – shodanshok Sep 18 '20 at 11:22
  • Lovely, that answers a lot of my questions. Just one last question about the media error count. When it says a small number like 1 that doesn't mean there is no more room in the spare sector, it still successfully remapped to a spare sector? And how is this number tracked, does the drive return the number of media error (glist) or is the RAID card that keeps count? So if I swap RAID card or if I clear the array and create a new one will it go back to 0? – Symbux Sep 18 '20 at 11:44
  • The disk will surely have enough space to remap some sectors, so one single error reported should not be fatal. However, be sure to check and monitor the disk state to detect any new, additional errors which would be a strong indicator of a disk going bad. The drive does not return the grown defect list (glist); rather the error is tracked by the RAID card. How this is done is implementation specific - it can be tracked in the persistent configuration storage of the NVRAM or, more probably, in the disk array itself (in a dedicated metadata area). – shodanshok Sep 18 '20 at 13:25

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.