(Disclaimer I'm no computer wiz) I had a disk drop into Error Mode.  It disappeared completely (additionally the SMART report simply said no such drive existed).  I was going to replace it thinking it was dead, but then realized it was plugged into a sata card (link) - and the only other drive plugged into that card was an unassigned drive.  I checked and sure enough it didn't show either.  I even unplugged a drive that was showing into the card and it showed up.  So I determined that the card was bad.  I replaced it and boom both drives (the one that dropped into error mode and the unassigned) show back up.  The one that dropped into error mode is now in process of rebuilding.  I did pull the diagnostics zip before I shut it down - and I have attached the SMART report from when I brought it back up (attached).

 

I was feeling pretty good about this until I thought... what if the card that has 4 drives attached to it dies???  Are all 4 drives going to go into error mode if that happens... will I just be up the creek at that point?  It is my understanding that Unraid can only rebuild one drive?  If 2 or more go into error mode is it game over?  Is there a way to emergency stop unraid if a disk disappears rather than try to write to it and cause it to go into error mode?  Again sorry for my limited understanding, but I'd appreciate some help or clarification here as I feel like I'm missing something or I would have been able to find something in the forums of this happening to someone else.   Thanks in advance!

unkevin-smart-20200902-1820-2.zip unkevin-diagnostics-20200902-1532.zip

1 hour ago, kpc180 said:

was feeling pretty good about this until I thought... what if the card that has 4 drives attached to it dies???  Are all 4 drives going to go into error mode if that happens... will I just be up the creek at that point?  It is my understanding that Unraid can only rebuild one drive?  If 2 or more go into error mode is it game over?  Is there a way to emergency stop unraid if a disk disappears rather than try to write to it and cause it to go into error mode?  Again sorry for my limited understanding, but I'd appreciate some help or clarification here as I feel like I'm missing something or I would have been able to find something in the forums of this happening to someone else.   Thanks in advance!

Should this ever happen to you, do NOT panic!   Take a deep breath and try to relax.  Then start a new thread and describe exactly what is going on.  Quite possibly, you have not lost any files.  They are perfectly safe on your hard drives.  If this is the case, the only way you can lose any files is through cockpit error!  Even if there are errors on a disk, you can recover the files on every other disk that was on that controller.  (Unraid writes files to its data disks using a standard Linux file format which means they can be read on any Linux computer.)  There are several Gurus who are willing to look at the problem and walk you through the steps to recover as much data as possible.  Each one of these cases are unique and advice will depend on the circumstances.

 

For this reason, I am not going to answer virtually any of your questions.  There is one point I will address.  Unraid will never write data to your Unraid array if the data can not be recovered through parity reconstruction. 

 

One more point.  If you have data that it truly irreplaceable, you should have another backup of that data somewhere-- Preferably offsite!   Disk failures are not the only way to lose data.  Fire, floods, hurricanes, tornadoes, lightning, theft of the server, and Ransomware are all rather common, if unlikely, occurrences.  (I can recall one Unraid user whose house was subjected to tidal flooding during a hurricane.  His server went partially under water.  He was able to recover the data from the disks that were not submerged.)

Edited September 3, 20205 yr by Frank1940

Thanks, that makes me feel a bit better.  But so I'm clear there isn't an option or something obvious I'm missing which would check to see if a disc went offline and then emergency stop unraid so that nothing tries to read/write to it and puts it in Error mode - and prevent a need for a rebuild?  

What I have is 'cold' precleared spare disk for each of my servers.  Then when one of the data disks goes off-line, I do the rebuild on the cold spare after pulling that disabled disk out. After that is done, I will wait a few days to see that everything continues to function normally.  Then I will run a couple of preclear cycles on the old data disk. It will usually pass. Then it becomes the cold spare. 

 

Rebuilding using the contents of the emulated disk to rebuild a physical disk has the advantage that if there was a 'bad' write to its data disk because of some problem external to the disk, the data on the emulated disk will probably be correct.  Yes, there is a way to force the array to accept that disabled disk as is and use it.  But it should be pretty far down on the list of options. 

 

My experience is that a disk going off line might happen once a year at the most.  If you spin your array disks down, it seems to extend their lives quite a bit. The three 1GB disks on my Test Bed server have an average age of over 8 years of power-on hours.  In the Media server, I have two remaining (out of four) Seagate ST3000DM001 drives (Which has a record of being the worst hard drive model for reliability of the 21st century) with more than seven years on power-on hours!

 

Make sure that you have the Notification system setup on your server to notify you of any problems.  I even have the "Array status notification:" turn on to send a daily e-mail to me telling everything is OK.  This also tests the server's mailing system to guarantee that I will get any E-mail notification of a more serious problem.