The best I can tell something attempted to write to one of my data drives about 4 days ago and couldn't reflecting two errors and now the drive is "red X'ed" / faulty.  At this point I'm wondering whether I should even attempt to salvage it in any way or just replace it and throw it away.  I'm not terribly good at understanding the diagnostics.zip data so the unRaid Wiki recommended to post something here.  The faulty drive appears to be sdh (WDC_WD20EADS, 2TB).  If the attached diagnostics point to any recommendations I would appreciate you passing them on to help.   

tower-diagnostics-20160801-0935.zip

Disk dropped offline so there's no SMART report, power down, check all cables, power back up and post new diags.

Thanks Johnnie.  I'll be able to do just that in a couple hours, but from the sounds of it this probably isn't a good sign unless it is just loose power cable.  I thought powering the drive down was possibly something unRaid did as precautionary measure if it found it couldn't write to the disk, but since you are asking for the SMART data it leads me to believe this isn't the case.  It just may be completely dead, which pretty much answers any questions for me.   I'll do as you recommended tonight and pass on what I find.  Thanks for the help.

I swapped the sata cable and power cable with another drive but the red x persists. I've attached the diagnostics. Is it simply dead?

tower-diagnostics-20160801-2028.zip

Your latest syslog has sdh as the cache disk. Unfortunately, saying something is sdh doesn't help identify it after a reboot.

 

There is this in your SMART:

Device Model:     WDC WD20EADS-00W4B0
Serial Number:    WD-WCAVY7298247

which is disk4. Is this the disk we're talking about? SMART for that drive looks OK. If you have another WD20EADS I don't see it.

 

The red X means unRAID has disabled the disk due to a write failure. After a write fails and unRAID disables it, unRAID will not use the disk again until it is rebuilt. This is because the data on the disk is no longer valid. unRAID is now emulating the disk using all other disks plus parity to calculate the data for the disk. The failed write was emulated, any subsequent writes to the disk were emulated, any reads from the disk are emulated. The valid data is in the parity array waiting for you to rebuild it.

 

What do I do if I get a red X next to a hard disk?

Thanks trurl.  Yes, it is the WD20EADS drive. Should I just put it through a preclear and then rebuilt it if unraid gives it a passing status?

Oh wait. I see reconstructing the drive just says to rebuild it without preclearing. I assume this safe if it has failed writes again...correct?

Oh wait. I see reconstructing the drive just says to rebuild it without preclearing. I assume this safe if it has failed writes again...correct?

Yes, just rebuild it.

 

Don't understand the question in the second sentence.

 

I rebuilt it and the disk went to "green" so I guess it is ok...for now.  I just did a parity check about two weeks ago.  Does it make sense to do one again since a data drive has been rebuilt?  Are there any other actions that might be recommended to do now that a rebuilt is complete?

 

PS - I do have plans to go ahead and buy another data drive and pre-clear it and just leave it sitting in the server.  I think it is only a matter of time before this data drive or another fails.

I rebuilt it and the disk went to "green" so I guess it is ok...for now.  I just did a parity check about two weeks ago.  Does it make sense to do one again since a data drive has been rebuilt?  Are there any other actions that might be recommended to do now that a rebuilt is complete?

 

PS - I do have plans to go ahead and buy another data drive and pre-clear it and just leave it sitting in the server.  I think it is only a matter of time before this data drive or another fails.

I always do a non-correcting parity check after a rebuild. That way parity is not affected. If you get zero parity errors then you know the rebuild was good.

I always do a non-correcting parity check after a rebuild. That way parity is not affected. If you get zero parity errors then you know the rebuild was good.

 

Really good advice.  Thanks for mentioning it.  I wouldn't have thought of it.  I'll do just that.

Parity check finished with no errors. I guess it's safe to use the server as normal... Correct?  The two write errors from before resulting in the faulty disk warning I guess are a good sign that is a matter of time before this drive fails.

Parity check finished with no errors. I guess it's safe to use the server as normal... Correct?  The two write errors from before resulting in the faulty disk warning I guess are a good sign that is a matter of time before this drive fails.

 

I would suggest that after a redball or redx you've changed the power and sata cables, done a rebuild, done a non error correcting Parity Check and there is nothing concerning in the SMART report then you should be fine with it. More than likely you've dealt with the culprit through cable replacement.

 

To be sure to be sure, run it through a long SMART test and give it another workout. If the disk comes though that ok I would say all is well.

Parity check finished with no errors. I guess it's safe to use the server as normal... Correct?  The two write errors from before resulting in the faulty disk warning I guess are a good sign that is a matter of time before this drive fails.

 

You already did what you should, swapped both cables, so if the same disks fails again in the near future replace it, sometimes a disk can be bad despite having a healthy SMART, but it's more likely that the disk is fine.

I am sorry to hijack this thread, but since this is a closely related topic, I will

 

When should we replace hdds? Obviously when they get redballed. Short of that, when we get yellow triangles and specific warning (reallocted sector count etc), should we replace immediately or just keep our eyes open and have a spare drive handy?

 

Thank you

I am sorry to hijack this thread, but since this is a closely related topic, I will

 

When should we replace hdds? Obviously when they get redballed. Short of that, when we get yellow triangles and specific warning (reallocted sector count etc), should we replace immediately or just keep our eyes open and have a spare drive handy?

 

Thank you

My general rules are:

• Any disk which has a "Failing Now" SMART attribute should be replaced ASAP
  
• Any disk that has a non-zero number of reallocated sectors where the value is not stable should be replaced ASAP.
  
• Any disk that has a non-zero value of reallocated sectors should be a replacement candidate although perhaps not urgently.  The larger the number the more urgent I consider the replacement
  
• Any disk that has a non-zero value of pending sectors should be removed from the array and tested ASAP.  Pending sectors are not necessarily a problem but a non-zero value affects the capability of rebuilding another disk without errors.  If after a preclear (or equivalent) the pending sectors goes to zero and the other SMART attributes still look OK then the disk can be consider as OK and suitable for reuse in the array.
  

There is also the fact that any disk that displays persistent issues and they follow the disk then it should be considered a replacement candidate even if SMART looks OK.

You were very clear and precise. Thank you!

 

sent from my mobile phone, i apologise for any spelling errors

 

 

I am sorry to hijack this thread, but since this is a closely related topic, I will

 

When should we replace hdds? Obviously when they get redballed. Short of that, when we get yellow triangles and specific warning (reallocted sector count etc), should we replace immediately or just keep our eyes open and have a spare drive handy?

 

Thank you

Actually, redballs (red X in v6) are more often caused by something other than the drive actually being bad, as seen in this very thread. I would say the majority are connection issues. All the redball (red X) means is that unRAID disabled the drive because a write failed. If the rest of the array is healthy unRAID is emulating the drive using the rest of the array.

 

Sometimes it might be worth rebuilding to a spare even if the drive hasn't been disabled yet.

 

A small number of reallocated sectors are probably OK if they don't continue to increase.

 

Pending sectors should be dealt with immediately but it might be possible to continue to use the drive.

 

Rebuilding to a spare if you have one is good practice even if the original drive isn't bad, since that gets your array back to normal and you can then take time to decide if the original drive is OK, clear pending, etc. But it is often not that difficult to decide the original drive is probably OK and you can just rebuild to it. I don't keep spares because I can just shut down if I have to and wait on Amazon Prime.

 

Some further reading for you:

 

Resolving a Pending Sector

 

Known ATA S.M.A.R.T. attributes

 

Thank you both for your help!

Following up on what @trurl wrote:

 

"All the redball (red X) means is that unRAID disabled the drive because a write failed"

 

Is there any possibility of this happening because of cpu/memory overload (eg too many dockers running or a docker utilizing as many resources as it could find available)?

 

It might be silly question, but I am still going to ask it

Thank you both for your help!

Following up on what @trurl wrote:

 

"All the redball (red X) means is that unRAID disabled the drive because a write failed"

 

Is there any possibility of this happening because of cpu/memory overload (eg too many dockers running or a docker utilizing as many resources as it could find available)?

 

It might be silly question, but I am still going to ask it

Impossible is a pretty big statement, but I would say extremely unlikely. A write failure is almost certainly a hardware issue of some kind, often just a loose connection.