0

My current NAS (Synology 8x10TB) is near capacity and I'm looking to buy another one (8x22TB) ASAP.

After reading an alarming amount of articles when purchasing the first one about how RAID 5 was obsolete, I opted for RAID 6, but with hard drive capacities more than double the original size, I'm even doubting the validity of RAID 6.

I was wondering what would happen when a URE was encountered during a RAID rebuild. I found another SuperUser question What happens when a RAID rebuild fails due to URE? which asks the same question, but the accepted answer states that the entire RAID is lost (and thus all contents). On the other hand, I often read that the RAID will still be rebuilt, but that certain blocks will be marked "unrecoverable", meaning that one would only lose an amount of files.

Which of the above is correct? Do you actually lose the entire RAID, or just an amount of files? And is the answer the same for all RAID controllers? If not, what would the answer be specifically for a Synology NAS?

On a side note: yes, I do understand that RAID is not a backup, all data on my Synology is backed up to another Synology, which in turn is backed up to separate hard disks. So I do have the backups, I'm just not looking forward to restoring any of them...

Improve this question
4
  • Does Synology handle ZFS? That has built-in resilvering that should avoid throwing out an entire disk just because of one URE
    – Chris Davies
    Commented Apr 3 at 21:02
  • @Chris Davies Synology does not handle ZFS I believe, however it does have BTRFS. I don't quite understand how this would help with a URE however... In case of a disk failure (or two failures with RAID 6), a URE on any of the remaining disks would still result in unrecoverable data, no? My guess would be that an unrecoverable block of data would also mean unrecoverable metadata (such as the data for the integrity check). Also, given the additionaly integrity data, would ZFS/BTRFS not use more diskspace for the same amount of data?
    – Deekay
    Commented Apr 3 at 21:30
  • Yes. The main problem is that a single read failure on one disk can throw out that whole disk even though the net result of reading that block from all disks means the data block is intact. Then the next URE - regardless of which disk it comes from - will throw out another disk with the result that your RAID5/6 crashes unhappily. The upside of ZFS is that, as I understand it, a single unreadable block doesn't fail the entire disk. I'm not familiar with BTRFS to know how it handles RAID-like configurations
    – Chris Davies
    Commented Apr 3 at 21:39
  • 1
    superuser.com/questions/1683865/…
    – Joep van Steen
    Commented Apr 4 at 0:26

0

Reset to default

You must log in to answer this question.

Browse other questions tagged .