[SOLVED] 2 disks have errors, one will fail soon, what to do now?

Reading some topics on swap-disable, it is clear that the procedure only works if my bad disk2 is actually red-balled (disabled), which it is not... now there are some tricks into making unraid think it is but i don't want to play around without ever having these procedures practiced...

 

So i think i will try to shrink the WD red first and see if that works.

Could it be the unraid webgui is showing the wrong size?

The gui shows 3,907,018,532 bytes for a 4TB Hitachi, while the SMART report shows 4,000,787,030,016.

 

Preclear also shows 4,000,787,030,016 for the WD red... which would mean it would fit... it's exactly the same size as the Hitachi it replaces...

Sounds like it may in fact be the same size ... in which case it should work fine as a replacement drive for the rebuild.

 

To confirm that, just shut down the system and install the WD Red;  then boot;  Stop the array and remove the failing drive from the configuration; then Start the array so it shows a "missing" drive; then Stop the array and assign the WD Red to that slot; and then Start the array again ... it should automatically do the rebuild.    If the drive is in fact too large, it will simply tell you that -- it won't hurt anything to try.

 

Ok, thanks, will try!

Ok, after preclear finished succesfully i've stopped the array, replaced the suspicous disk2 with the new WD red. After restarting and assigning the WD red to disk2' slot, it said ' Wrong disk' with a red ball. Not totally what i expected according to the wiki, but i could choose to rebuild, and i continued. It is now rebuilding... fingers crossed

It's working fine -- that's exactly what it should show.

 

If it had a problem with the size, it wouldn't have started.    Note also that the disks are shown as the exact same size.    Not sure why you thought there was a difference, but the details on the Web GUI show exactly the same byte count for all nine of your array disks (parity plus the 8 data drives).

 

No, there is something wrong with the way the webgui shows the size.

Now the WD is part of the array, it shows the same size as the Hitachi's (3,907,018,532).

 

But when i was preclearing it, the webgui showed the WD as 3,907,018,584...

The SMART report for both drives showed 4,000,787,030,016....

 

Also, i would expect the array to in a stopped state while rebuilding? It looks like it just started. I'm not gonna touch it until it is finished...

 

(97% rebuild and counting... )

Ok, rebuild was done in 8 hours, looks ok...

Ok, rebuild was done in 8 hours, looks ok...

I'd get new smart reports on all the drives again, and compare.

Does unraid save all SMART reports somewhere? Otherwise i have nothing to compare to...

No, there is something wrong with the way the webgui shows the size.

Now the WD is part of the array, it shows the same size as the Hitachi's (3,907,018,532).

 

But when i was preclearing it, the webgui showed the WD as 3,907,018,584...

The SMART report for both drives showed 4,000,787,030,016....

 

There's nothing wrong with the sizes.  SMART is displaying the actual byte count in decimal.  The Web GUI is displaying it in KB.    Remember that disk drive manufacturers report sizes in decimal;  computers "count" in binary, which 1KB = 1024 bytes.

 

Note that 4,000,787,030,016 / 1024 = 3,907,018,584  ... which is consistent with what's displayed (the 52KB difference is because the actual partition doesn't use the entire disk to the GPT partition tables.

 

Does unraid save all SMART reports somewhere? Otherwise i have nothing to compare to...

Base unraid does not. There were (are?) add ons that did, but if you had installed them, you would know.

 

It's just a good idea to periodically collect and save smart reports on all your drives, that way you may get an early warning that you need to deal with something.

Ok, i'll run reports on al disks and save them so i can compare later on.

 

New disk seems ok. I understand the technique behind parity and rebuild, but still... it is fascinating to see it actually works

 

Should i do a new parity check ny the way? The unraid gui now says there has never been one.

Should i do a new parity check ny the way? The unraid gui now says there has never been one.

Definitely, and it should be a non-correcting check. The drive rebuild basically did a blind write to the new drive, there was no verification that the written data actually was committed correctly. Chances are everything is OK, but you want to find out now if something is amiss, so you can pull the drive and try the rebuild again. If the rebuild wasn't successful, the parity check will show errors, and you don't want the errors written to the parity drive.

Agree ... after a rebuild is the only time I ever suggest using a non-correcting check; but it IS the one time you want to do that, so if anything went awry with the rebuild you can do it again.

 

Thanks. Non correcting parity check is started.

Would be nice if this step was mentioned in the wiki...

Thanks. Non correcting parity check is started.

Would be nice if this step was mentioned in the wiki...

 

It is but in the user contributed manual.

Good news the failed Hitachi 4TB (disk2) is suited for RMA It's under guarantee until may 2015.

So, with a bit of luck i will get that one replaced.

Parity was OK.

 

Now, i've still got my disk6 issue. It has not yet failed, but the reallocated sector count is not zero either. See smart report ive posted. I've copied all data off it, so it is empty now, and the count hasn't increased in the meantime.

 

It is under guarantee until august 2015... but, for it to be RMA, it has to fail, which it has not. What should i do, use it with some disposable data until it does fail and RMA it? Is there a way to speed up failing ?

I would RMA it anyhow. I don't think they test all of the disks returned as I had a similar issue with a WD drive and they still replaced it. Once you start getting reallocated sector counts its only a matter of time and if you carry on using it your more at risk of it and another disk failing at the same time.

r.e. Disk 6 =>  You could RMA it (as noted above, I've never seen an RMA refused);  or you could simply replace it and use it as a backup.    Many of my backup disks are drives that were either replaced with larger ones, or that had minor issues (like reallocated sectors) and I didn't want to keep them in active service.    Note that if you simply rewrite all of the sectors on the disk, any pending reallocations will be done; and reallocated sectors themselves aren't "bad" -- the drives are in fact designed for that.  Since backups aren't used a lot, it's an excellent use for those drives.

 

I've never RMA'd a disk, so i think i will RMA the failed disk2 first, if i get a replacement, i'll use that to replace disk6.

Then i will try to RMA disk 6, and it's replacement (if i get it RMA'd) will be added to the array, or will be prepped as a spare.

 

In the meantime i will monitor disk6 closely.

... if i get a replacement, ...

 

??  There's no doubt you'll get a replacement 

... Note that it's likely to be a reconditioned/recertified drive -- but it will still be warranted for the remainder of your original drive's warranty.

 

I always use drive I receive as RMA replacements as spares, or as extra storage for one of my desktops -- I never add them to my UnRAID servers => I simply want new, fully warranted drives for my storage arrays.    But that's your choice ... the drive will indeed be fine -- it's just not going to have much of a warranty left.

 

 

 

After a single pre-clear cycle the pending count should go to zero and the reallocated count may or may not increase. If the drive passes a couple of pre-clear cycles, the pending count goes to zero and stays there, and the reallocated count stays stable then use the disk.

@garycase i did not know they replace it with a reconditioned/efurbished one with limited guarantee, good to know.

@dgaschk definitly gonna try that, good info