NAS HDD S.M.A.R.T "Bad"

[sal]

If the content availability is not critical why waste drive space with RAID, just mirror the server to a second one which provides proper physical redundancy against fire, theft, flood, providing they're not at the exact same physical location. i.e. one could be in a secure outbuilding or nearby relative, depending on how painless you want the syncing to be.

One of those cheapo HP Microservers as mentioned by others would provide a nice second solution acting as primary while you replace each 'failing' disk under warranty with no down time.

So your solution to "wasting" 1xHDD for RAID 5 array for a non-critical system is to waste 4xHDDs + compute node + running cost?

[Hobosapien]

If the content availability is not critical why waste drive space with RAID, just mirror the server to a second one which provides proper physical redundancy against fire, theft, flood, providing they're not at the exact same physical location. i.e. one could be in a secure outbuilding or nearby relative, depending on how painless you want the syncing to be.

One of those cheapo HP Microservers as mentioned by others would provide a nice second solution acting as primary while you replace each 'failing' disk under warranty with no down time.

[eek]

Building something from scratch is not an option, it's just a media server so nothing important enough to throw real money at it.

HPE ProLiant Gen8 4GB RAM MicroServer | Ebuyer.com £125
Freenas = Free
or unraid but this is ott for what you need.

[sal]

RAID 6 is not your best option in 4 disk RAID as it "wastes" 2 disks for parity which is half in this case. Meaning that RAID 10 is the best option if you want 2 disk resilience as the performance is much better than RAID 6. This is not the case with 4+ disk, like eek's 12 disk array, where you only "lose" 16% RAW space with RAID 6 vs 50% with RAID 10

Nothing wrong with using RAID 5 for not mission critical data like media files in home server. Just be extra cautious during the rebuild giving the fact that the first time it will need to run with one disk already going bad.

If you are planning on buying a new set of larger disks, you might as well build a new system for as little as £100 added cost on top of the price of the HDDs, which will allow you to keep using the old 2TB disks in the existing system.

[SimonMac]

Building something from scratch is not an option, it's just a media server so nothing important enough to throw real money at it.

The options the NAS has is RAID 0,1,5 and 10, by the looks of it RAID 6 would be the best option for me, however the options limited to those means RAID 5 if I want more storage but only 1 disk tolerance, or RAID 10 for two disk tolerance but limited to half capacity?

I have 4 x 2TB in RAID5 which gives 6TB useable, as this is running low I could get away with 4 x 3TB in a RAID 5 configuration for 9TB useable, or would need 4 x 4TB in RAID 10 to give me an increase in useable storage and extra fault tolerance?

[eek]

With Raid 5 you can only remove/lose 1 disk at a time without losing all the data. If S.M.A.R.T. sensors are reporting the disk as bad this should be replaceable under warranty, even though the drive is not failed yet. Unfortunately you will have to go through the process twice, waiting for a full rebuild of the array after the first disk replacement.

I would suggest backing up all the critical data on another system/cloud before proceeding as you are in danger of encountering the biggest flaw of RAID 5 in the 1TB+ HDD era - second disk failure during rebuild after one disk failure. As the rebuilt process buts a lot of stress on the array and one of the remaining disks is already reported as bad, this scenario is not far fetched.

eek's suggestion of double parity is pointless in 4 disk array as if you are going to lose 1/2 of the RAW space you are better off with RAID 10 as the performance is much better. (unless that's what he meant in the first place, although this is not a parity)

I didn't suggest it - I pointed out that I have a system with dual parity (its 27TB across 12 disks excluding the parity disks for reference so rather larger than SimonMac's)...

I personally have always avoided Raid 5 is an all or nothing approach is the reason why I have always built my own server using either snapraid or lime-technologies' unraid to manage parity disks...

In my worst case scenario I now need to lose 3 disks before I start losing data - and even then I will only lose the data on the disks that are lost...

[sal]

With Raid 5 you can only remove/lose 1 disk at a time without losing all the data. If S.M.A.R.T. sensors are reporting the disk as bad this should be replaceable under warranty, even though the drive is not failed yet. Unfortunately you will have to go through the process twice, waiting for a full rebuild of the array after the first disk replacement.

I would suggest backing up all the critical data on another system/cloud before proceeding as you are in danger of encountering the biggest flaw of RAID 5 in the 1TB+ HDD era - second disk failure during rebuild after one disk failure. As the rebuilt process buts a lot of stress on the array and one of the remaining disks is already reported as bad, this scenario is not far fetched.

eek's suggestion of double parity is pointless in 4 disk array as if you are going to lose 1/2 of the RAW space you are better off with RAID 10 as the performance is much better. (unless that's what he meant in the first place, although this is not a parity)

[eek]

I have a WD DL4100, inside is 4 x 2TB WD Red HDD configured as RAID5, however two are now showing as "bad", what is the best way to get this resolved?

The drives are still under warranty so I could send two drives away and wait for them to replace them, but how much of a risk is that to the NAS running on only two drives in the meantime?

And does a BAD SMART reading always mean the drive is bad, in other words can I send it back to WD and they say the drive is fine?

Well 2 dodgy disks is a problem as you can only replace 1 at a time without losing data - which is why I use a system with dual parity disks.

[SimonMac]

I have a WD DL4100, inside is 4 x 2TB WD Red HDD configured as RAID5, however two are now showing as "bad", what is the best way to get this resolved?

The drives are still under warranty so I could send two drives away and wait for them to replace them, but how much of a risk is that to the NAS running on only two drives in the meantime?

And does a BAD SMART reading always mean the drive is bad, in other words can I send it back to WD and they say the drive is fine?