Hi guys. So I check my unraid system home page every few days and this morning I noticed that I have a red x at a data drive. The data drive also has 1850 errors and a huge amount of writes. Also worryingly I have 17 errors on my parity drive too but it's still on a green light. I currently have no spare drives so I will need to order one asap. I preformed parity checks and smart checks recently and all was ok. My parity drive is a 6tb western digital red, and 2 data drives are Hitachi nas 4tb with the suspect drive being a seagate nas 4tb. Its about 2 years old.

 

I tried to spin up the faulty drive to get a smart report but it won't spin up remotely. I'm at work and have vpn ed into my home network so I can't actually goto the server to check it out.  I have attached a log I got from the server. I'm on a smartphone doing this so am a bit limited. At this time I'm just looking for a direction of travel to go in.

 

I think I will order another drive. I was considering a 8tb seagate archive drive and swapping out my parity drive but I don't think I can do that at the moment. Also seagate looks to be a problem as I have had 2 faulty seagate cache drives previously.  I do have another sata drive bay in my server.

 

I do have a usb 2 tb drive and I could free up some space in other machines so was thinking I could transfers data from the failed drive and clear it that way? Then remove for some testing? Cheers des

tower-diagnostics-20151125-0855.zip

A drive being 'red-balled' means that a write to it failed for some reason.  Although this can indicate a genuine problem with the drive as often as not the drive is fine and it is an external factor such as a loose SATA/Power cable that caused the problem.  Things like vibration can cause temporary cabling issues even on systems that have been stable for ages.  At this point unRAID stops writing to the drive until appropriate recovery action is taken.

 

You were not able to get a SMART report for the problem drive.  This suggests that it has currently dropped offline.  Typically rebooting the system brings the drive back again if it has not physically failed.  A SMART report for the drive is really required to see if it genuinely has issues, or if the problem was likely caused by an external factor.  You may want to reseat any cables just in case that was the issue.

 

If the drive turns out to be OK, then you could recover by rebuilding the drive back onto itself rather than replacing it.  Having said that if a spare drive is available then rebuilding onto a spare is better from a recovery point of view as the original drive is still intact for recovery purposes.  It is highly likely that all the data on it is still intact and recoverable.

 

BTW: 

I have a good experience with using the Seagate 8TB archive drives as both parity and data disks with unRAID.  I now buy these where previously I used WD 6TB Red drives as the 8TB is typically cheaper so offers better value for money.  I think other unRAID users have reported good experiences with these drives.

Thanks for the help. So I will try resetting the server when I get home.  I will keep you updated. I can get a 8tb seagate for £160 and a 6tb red wd for £170. Choices choices. But if my 4tb seagate is dead is there a way to upgrade by parity to a new 8tb drive without losing data? I would then use my current 6tb parity as a data drive?

Thanks for the help. So I will try resetting the server when I get home.  I will keep you updated. I can get a 8tb seagate for £160 and a 6tb red wd for £170. Choices choices. But if my 4tb seagate is dead is there a way to upgrade by parity to a new 8tb drive without losing data? I would then use my current 6tb parity as a data drive?

There is a process called "Swap-Disable" where you simultaneously upgrade parity and use the old parity disk to replace an old data disk.  This might well be an appropriate time to use it.

 

The process works in several stages where the first stage copies the old parity disk contents to the new parity disk, and then starts a rebuild of the replaced data disk onto what was previously the parity disk.

Sounds like what I would want. Any links to how to do that? Do I need to set the new drive to zeros?

 

Sounds like what I would want. Any links to how to do that?

Do not have this to hand but it should be easy enough to find.

Do I need to set the new drive to zeros?

Not strictly necessary for the swap-disable, although you might want to put any new drive through a preclear cycle as a confidence test of the drive.  Having said that the process itself is not a bad confidence check if it completes without error.

ITIMPI thanks for the help so far.  Right i ordered a 8TB Seagate archive drive at lunch time so should have it by this time next week.  I will then do a single preclear (might do 3 if my system looks ok) and fit it as parity. 

 

Now im home i restarted the machine and re seated the drives, and can now access the red balled drive and do a smart test.  The results look ok, to me at least.  The drive is still red balled so i need to some how rebuild it.  So im thinking of stopping the array, removing the drive from the assigned list and then re adding it?  Does that sound correct.  Then it should be rebuilt by unraid?

 

Here is my new fault logs.

 

 

 

Thanks Des.

tower-diagnostics-20151125-2009.zip

Ok i found this

 

https://lime-technology.com/wiki/index.php/Replacing_a_Data_Drive

 

The drive is rebuilding now.  I will report back if i have any issues. 

ITIMPI thanks for the help so far.  Right i ordered a 8TB Seagate archive drive at lunch time so should have it by this time next week.  I will then do a single preclear (might do 3 if my system looks ok) and fit it as parity. 

Fair enough, but be aware that with a drive that large each preclear cycle can take a couple of days and you may not want to wait that long.  In practise assigning the drive as a replacement parity drive is probably as good as doing a preclear as it involves rebuilding parity from the data drives and writing to every sector (but keep the old parity drive until it parity has successfully been built  in case something goes wrong).

 

Now im home i restarted the machine and re seated the drives, and can now access the red balled drive and do a smart test.  The results look ok, to me at least.  The drive is still red balled so i need to some how rebuild it.  So im thinking of stopping the array, removing the drive from the assigned list and then re adding it?  Does that sound correct.  Then it should be rebuilt by unraid?

I see you seem to have found the procedure for forcing a drive to rebuild onto itself.  Hopefully this will succeed without issue.

Ok the drive rebuilt itself overnight without issue. I'm doing a parity check now to make sure all is good. Cheers des

Ok guys an update to this. I received my 8tb seagate and have been running 3 passes of preclear on it. It took a week. Anyhow literally the day the last preclear was finishing I got a red ball on the 4 tb seagate again. It's on a different slot in my microserver than it was the first time it red balled. So anyhow I set unraid to rebuild the data on the 4 tb seagate again and it completed that overnight. I then swapped the config to drop my 6tb Wd red as parity and use my 8tb seagate as parity. So parity is now being built on the new 8tb seagate parity drive. The 6tb old parity drive is still in the machine as unassigned.

 

Moving forward once the parity is complete on the 8tb seagate I want to copy what's on the 4tb seagate drive to the old parity 6tb Wd drive and remove the 4 tb seagate nas drive for more testing.  To do this I'm thinking I need to zero the 6tb Wd old parity drive and install it in the array. Then format it as zfs. Then somehow copy or move everything from the 4tb seagate into the 6tb Wd drive. Then remove the seagate 4tb. Does that make sense? How should I go about that? Also I wanted to use the new zfs format as I heard the old riserfs can give problems?

 

Thanks des.

...

Moving forward once the parity is complete on the 8tb seagate I want to copy what's on the 4tb seagate drive to the old parity 6tb Wd drive and remove the 4 tb seagate nas drive for more testing.  To do this I'm thinking I need to zero the 6tb Wd old parity drive and install it in the array. Then format it as zfs. Then somehow copy or move everything from the 4tb seagate into the 6tb Wd drive. Then remove the seagate 4tb. Does that make sense? How should I go about that? Also I wanted to use the new zfs format as I heard the old riserfs can give problems?

 

Thanks des.

Perhaps you meant xfs instead of zfs? Don't attempt to format it yourself, let unRAID format it when it is ready to.

 

Preclear the old parity and add it to a new slot. When you start the array, unRAID will offer to format it. XFS is the default for new disks. Then you can copy from the 4TB. To remove the 4TB you will have to New Config, reassign all your drives without the 4TB, and rebuild parity. Be very sure you don't accidentally assign a data drive to the parity slot or unRAID will overwrite your data with parity. If you have any questions about any of this be sure to ask.

 

There is a sticky in this subforum all about moving data around to change filesystems.

Thanks for your response. Yep I ment xfs. Not sure were zfs came from. So I'm finished with the parity sync on the new 8tb seagate parity and I'm almost finished the preclear or the 6tb Wd drive that was my old parity. Next step is to add the 6tb Wd to the array as a new data drive. Then copy across the data from the flakey 4tb seagate to new new drive. I believe this is this sticky you mentioned I need to follow.  Post 3?

https://lime-technology.com/forum/index.php?topic=37490.0

 

So when I copy the data across I then remove the flakey seagate and create a new config being sure to set the drives up correctly. I will take a picture of the screen with the drive serial numbers on to be sure. Once the flakey seagate is removed I need to test it hard to find out what's going on. Would you recommend preclear or something else to do that? At the moment smart looks ok with no reallocated or pending sectors. When it fails it takes a reboot to get it to even show in unraid. Is that a typical failure mechanism? It's under warranty so I want to kill or cure it asap.

 

Thanks des.

 

 

Yet another update. So the darn 4tb seagate has red balled again ?. At this point I'm thinking of just having unraid rebuild onto my 6tb Wd drive which has almost finished zeroing in preclear, and forgetting about changing to xfs. Does that sound like a good plan? Is xfs an important upgrade?

Ok so im rebuilding my data onto the 6tb wd drive and have removed the flaky seagate 4tb from the array.  Now i have restarted the machine the seagate is showing as only 4.14GB not 4TB!!  But in the preclear plugin its showing as 4TB.  Also the serial numbers are slightly wrong.  WTF? 

 

Please find attached screen shots of the SAME DRIVE!

 

And the second screen dump.

 

What should i do now?  Thinking of trying to run 3 preclears on it?

 

Cheers Des

And the latest diagnostic logs.  Smart still looks ok to me..

 

tower-diagnostics-20151214-1824.zip

So the rebuild finished and I'm doing a parity check on the system now. The seagate drive thats flakey is in another Windows box and is doing some seagate seatools tests. But unfortunately it's smart info looks good when it's accessible. As I seem to need a Fail report from seatools and the fault is intermittent I'm not sure what to do. I forced the fault by rotating the drive from the horizontal position to vertical with it running. I wasn't violent with it and didn't bang it.

 

Any ideas on how to get this to properly fail? Some Google searching seemed to reckon that failed drives can show up as 4.14 gb capacity. Odd.