HELP. One drive dead

[pras1011]

Hi,

 

I was doing a parity check and disk 2 went offline. Currently the server activity light on disk 2 is still on.

 

But I have shut down the server just in case.

 

I don't have a spare hdd but please could you check my syslog and tell me what the next steps are?

unraid-syslog-20180127-2114.zip

[JorgeB]

10 minutes ago, pras1011 said:

But I have shut down the server just in case.

Power the server back on and please post your diagnostics: Tools -> Diagnostics

[pras1011]

Please see attached.

unraid-diagnostics-20180127-2314.zip

[JorgeB]

The failed disk is missing, not a good sign after a power cycle, you need to replace or swap cables with another disk to rule them out, if still missing the disk might be dead, if it's detected post new diags so we can see a SMART report.

[pras1011]

So to reactive disk 2 I need to stop the array and reassign the disk? What’s the best way to get smart data? I will try the cable replacement first though.


Sent from my iPhone using Tapatalk

[JorgeB]

The disk isn't being detected, first you need to get it online, if the disk isn't dead.

[pras1011]

Ok. So what I will do is to put disk2 in an external enclosure and see if its detected from there. If detected and there is a smart report, I will start changing cables.

[pras1011]

This is the smart report from disk 2.

[JorgeB]

That's not a good sign but not enough of a reason for the disk not being detected by unRAID, still, and if it's under warranty you might want to get it replaced.

[pras1011]

I put the disc back into unraid and it showed up. However unraid it warning of the reallocated sector count. All other hdds are zero but disk2 is 11728.

 

Are you sure that unraid wont disable a drive because of this?

 

My biggest concern is whether the data on disk2 was written to properly when I was writing to it (before the failure)?

[JorgeB]

Are you sure that unraid wont disable a drive because of this?

It was likely disabled because of this, it just doesn't explain why it wasn't detected after that.

 

My biggest concern is whether the data on disk2 was written to properly when I was writing to it (before the failure)?

In theory it should be all in the emulated disk, in practice it's possible that the file that was being written at the time the disk switched to being emulated has some corruption, but all the rest would be OK.

 

 

[pras1011]

But at the time of the failure, the hdd was only being read due to the parity check and no writing was taking place.

[JorgeB]

24 minutes ago, pras1011 said:

But at the time of the failure, the hdd was only being read due to the parity check and no writing was taking place.

Then you have nothing to worry about, I only mentioned that because you said:

19 hours ago, pras1011 said:

My biggest concern is whether the data on disk2 was written to properly when I was writing to it (before the failure)

I understood that as the disk failed while you were writing to it.

[trurl]

If a disk read fails, unRAID may read its data from the parity calculation instead, and then try to write it back to the disk. When that write fails, the disk is disabled.

 

It is usually possible to get the SMART from unRAID instead of putting the disk into another system. Note that it is almost always a bad idea to put an unRAID disk into another system and try to work on it there, since you will invalidate parity.

[pras1011]

Yes, in the syslog there were some reads and at the end there were some writes before the disk was disabled. Anyway, I have sent the disk back to Seagate. Hopefully it will be a quick turnaround.

[JorgeB]

2 hours ago, pras1011 said:

Yes, in the syslog there were some reads and at the end there were some writes before the disk was disabled

That's normal, when there's a read error unRAID uses parity plus all other disks to compute the correct data and tries to write it back, if the write fails the disk is disabled.

[pras1011]

Is there a way to run preclear without unraid? I want the server to to be unpowered and all hdds spun down as I am worried about other hdds failing whilst the server unraid is prelcearing the new hdd.


Sent from my iPhone using Tapatalk

[JorgeB]

You don't need to clear or preclear a replacement disk, but if you to run it to burn in a disk you can use an unRAID trial key in any other available computer.

[pras1011]

If I install an non-clear/non-pre cleared disk then unraid will clear is first and then rebuild the data? How long does the clearing process take on a 8tb disk?


Sent from my iPhone using Tapatalk

[JorgeB]

30 minutes ago, pras1011 said:

If I install an non-clear/non-pre cleared disk then unraid will clear is first and then rebuild the data?

No, it will start rebuilding immediately, a disk is only cleared when it's a new disk added to the parity protected array.

[pras1011]

I see. Interesting. Then what’s a good alternative preclearing program that I can run on Windows.


Sent from my iPhone using Tapatalk

[JorgeB]

There's no preclearing alternative, there are a lot of disk testing alternatives, you can use for example the manufacturer tools, like SeaTools or WD Data Lifeguard

[SSD]

It is very possible to boot unRaid on a Windows computer and run preclear. No need to define an array, and disks will not be corrupted in any way if you never define or start the array. Afterwards you can remove the precleared disk and reboot into Windows. If you have an old computer this may be a good option.

 

Preclear does two things. It clears the disk which is only needed to quickly add a new disk to a parity protected array. And it tests the disk pretty thoroughly by prereading the entire disk, then filling the disk with zeroes, and finally verifying the disk is zeroed. This action helps the SMART system in the drives to identify and mark bad and marginal sectors.

 

Note that unRaid's clear function only zeroes the disk. The preread and post validation reads are not done. Also, with unRaid, the disk is added to the array immediately. With preclear you look at the results and possible SMART attribute issues after the preclear and then decide whether to use the drive or perhaps return for an exchange if it has issues.

[pras1011]

I don’t have a spare computer but there must be a way to run preclear in a Windows environment?


Sent from my iPhone using Tapatalk

[trurl]

1 hour ago, pras1011 said:

I don’t have a spare computer but there must be a way to run preclear in a Windows environment?

No. But if you want to test the disk you can use the disk manufacturer's utility in Windows as already mentioned.

 

Since your intention is to rebuild an existing disk you don't need a clear disk. The only time unRAID requires a clear disk is when ADDING a disk to a NEW slot in an array that already has valid parity. This is so parity will remain valid.

 

I know I'm sort of repeating what has already been said, but sometimes different ways of saying things helps with understanding.