Drive disabled - should I replace it?

[moleboy]

Hi there,

 

I'm a bit of a noob with Unraid so forgive me whilst I am learning the ropes.

 

I have version 6.7.2 installed and I've just noticed whilst downloading some Media that my main (largest 8TB) drive (I only have 3) has been disabled due to write errors.  Whilst performing a preclear of the drive I was warned of errors then but thought it was a dodgy SATA cable.  The errors appear to be back.  I am attaching the diagnostics I got before rebooting.  Could someone take a look and tell me if they think the drive needs replacing?  There are no errors on the parity or the other 2 smaller drives.

 

I didn't have a SMART report for the disabled drive until I rebooted.  I ran a short SMART test (is that sufficient or should I run an extended?) and it came back error free.  I have attached this too.

 

Thanks for your time

 

 

mothership-diagnostics-20191218-2133.zip mothership-smart-20191218-2217.zip

[trurl]

Looks like CRC errors, which is typically a connection issue. Check all connections, power and SATA, both ends including power splitters.

 

Do you know how to rebuild the disk to itself?

[moleboy]

Hi,

 

Thanks for your response.  I will open up the case and recheck all connections.  

 

No I'm not too sure on how to rebuild the disk to itself.  Would this delete all the data off it?  When I try starting the array the drive is still disabled despite me rebooting and running a short Smart Test?

[trurl]

The drive will be disabled until it is rebuilt. Since a write to it failed, it is out of sync with parity. But the data for that failed write is still in the array from the parity calculation. In fact, Unraid will not even use a disk after it has been disabled. All attempts to access the disk, whether reading or writing, is done by emulating the disk from parity. So, the contents of the disk can still be read, but that is accomplished by reading parity PLUS ALL other disks and calculating the data for the disk. And all writes to the disk don't actually write the disabled disk, it just updates parity as if the disk were written.

 

So, the disk must be rebuilt. Rebuilding will not delete all of the data off it. It will instead write all of the data onto it, including the failed write and any subsequent writes that didn't actually go to the disk because it was disabled.

 

The safest approach is to rebuild to a new disk, since that allows you to keep the original disk as a backup in case there is some issue during rebuild. But obviously that requires a spare disk.

 

The most likely cause of an issue during rebuild is another bad connection, possibly with another disk. All disks must be reliably read in order to calculate the data for the rebuild. When I said check all connections, I meant all, even those for other disks. It is very common for someone to disturb the connections of other disks when they are inside the case.

 

SMART for that disk looks fine except for the connection issue so if that is fixed it should be OK to reuse.

 

To rebuild the disk to itself:

1. Stop array
2. Unassign disabled disk
3. Start array so the missing disk is registered
4. Stop array
5. Reassign disabled disk
6. Start array to begin rebuild

 

 

[moleboy]

Hi Trurl,

 

Thanks very much for taking the time to explain all this to me.  I'm just back on this now (bit of a time difference with me being in the UK) and will follow your steps and see how I go.  It would be comforting to know the drive isn't fried as I can't afford another 8TB disk at the moment so hopefully the rebuild will work out okay.  I replaced the SATA lead on the drive last time I spotted errors so I'm wondering if it is the controller port.

 

I'll check the other drive connections and switch the lead to another port and let you know how I get on.

 

I really appreciate your clear instructions!