I keep struggling with the same issue, which makes Unraid basically unsuable for me.

 

I have created a new array of 16 disks including 1 parity. Everything works well for a week or so, but then one disk shows "failed". Yet again, the disk is not faulty and it is a different disk every time.

 

I have now updated all firmware to rule out any related issues.

 

Any other idea what could be driving this? Diagnostic attached. Thanks!!!

tower-diagnostics-20151107-0413.zip

This is sounding eerily like the exact issue I've been fighting this week.

 

I upgraded my 4TB parity with a 5TB last week, and the array rebuilt parity just fine. I then replaced a 1.5TB with the old 4TB parity, and during data rebuild about 4-5 hours in, the 4TB dropped out, and several other drives as well. Rebooted, tried again, same thing, same time frame. I tried putting the 1.5TB back in, but it wouldn't take it (too small now).  I got a new 5TB, and during rebuild several drives dropped after a few hours. I rechecked all cable connections, reseated pci cards, and now I've upgraded from 650w to 850w single 12V rail PSU.  Started another rebuild, drive failed.....

 

 

@glave, any diagnostic log?

 

I just rebuild and this time it was stable for two weels. Now yet again a disk failed / disabled, which I yet again suspect to be working fine. Anyone can please have a look at updated log below. Thanks!

tower-diagnostics-20151123-2226.zip

Any thoughts based on my diagnostic? It is really annoying that I canoot get Unraid to work for more than a few days before it ejecting a disk... Thanks for looking into this!

Looking at your logs, errors occurred in different disks but all were connect to the M1015, you may have a bad controller, trying a different one would be my first step.

 

Unrelated to your issue, your last log is filled with these errors:

 

Tower kernel: read_file: error 2 opening /boot/config/super.dat
Tower kernel: md: could not read superblock from /boot/config/super.dat

 

You should remove and run chkdsk on the usb flash.

 

I have bought a new M1015 to eliminiate this as a source of error.

 

After replacing the M1015, things actually worked for a few weeks, so I was very excited that the error has been solved and I can now finally use Unraid.

 

Unfortunately, a drive just failed again. Would be great if you can take a look at the log whether this is yet again some controller issue, this time actually a really failed drive, or any other reason causing this?

 

Big thanks in advance!

tower-diagnostics-20160130-0907.zip

Syslog shows disk2 was disabled but SMART for that disk looks OK. Bad connection?

Thanks for your quick reply. Do you think this is the same "controller issue" I had many times before? Or something around a broken cable? Shall I just rebuild and give it another try? Anything else I can do. Restarting would not be sufficient to solve things, is it? Once a drive is disabled, I cannot enable it again, can I?

The disk is disabled because its data is no longer valid. The failed writes are in the parity array and need to be rebuilt to the disk.

Got it, I can rebuild to a new disk. Thanks.

 

However, not clear to me what has been causing the issue. Since starting to use Unraid, I never got it working properly. Disks always disable after some time. It used to be a few days with my old M1015. i replaced it and thought it was solved, but now yet again a disk disabled. If the disk is broken, this is great. But I read from your reply that the disk is fine. What else could cause Unraid to dusable disks and how can I get Unraid to work for me?

 

Thanks again for your help!

These types of issues are almost always due to bad or loose cabling. SATA connections are very finicky and a connection that is not flush and perfectly square is likely to cause problems with the pounding unRaid gives it. People often ignore this answer or treat it lightly, but it is true. Locking cables are excellent at resolving this issue!

 

Replacing a power supply is a major event in the server and often all of the cabling is unplugged and replugged in the process, or at least everything gets a good check, which actually is the reason changing the power supply often fixes the problem.

 

You can use either method.

These types of issues are almost always due to bad or loose cabling. SATA connections are very finicky and a connection that is not flush and perfectly square is likely to cause problems with the pounding unRaid gives it. People often ignore this answer or treat it lightly, but it is true. Living cables are excellent at resolving this issue!

 

Replacing a power supply is a major event in the server and often all of the cabling is unplugged and replugged in the process, or at least everything gets a good check, which actually is the reason changing the power supply often fixes the problem.

 

You can use either method.

 

Got it, this is very helpful. So, you are also confirming that it is a different issue than I used to have with my previous M1015. Someone in this thread had replied with my previous issue that this was not caused by a cable as the error message would be different. So, with the new issue, this is much more likely to be the cable.

 

Let me buy a new cable to fix the issue. What do you mean above by "living cable"?

Locking cable. Swype keyboard not perfect!

Thanks for your help.

 

Wanted to pick your brains whether it is safe to re-able the drive (https://lime-technology.com/wiki/index.php/Troubleshooting#Re-enable_the_drive) rather than to replace it (https://lime-technology.com/wiki/index.php/Replacing_a_Data_Drive).

 

Also more question if I have to replace it. Does this require a formatted new disk or can I just take a stock-disk. I don't have a precleared disk now, but thinking to do so for the future. I could not find an instruction though how to preclear.

 

Thanks in advance for your help!

Thanks for your help.

 

Wanted to pick your brains whether it is safe to re-able the drive (https://lime-technology.com/wiki/index.php/Troubleshooting#Re-enable_the_drive) rather than to replace it (https://lime-technology.com/wiki/index.php/Replacing_a_Data_Drive).

 

Also more question if I have to replace it. Does this require a formatted new disk or can I just take a stock-disk. I don't have a precleared disk now, but thinking to do so for the future. I could not find an instruction though how to preclear.

 

Thanks in advance for your help!

Never use the word format in any discussion involving rebuilding a data disk!

 

Formatting is never part of a rebuild. Format means "create an empty filesystem" and you definitely don't want to do that while rebuilding a disk.

 

You can use any disk to rebuild as long as it is at least as large as the disk you are rebuilding, but not larger than parity. And of course it must be a good disk that can be trusted. The point of preclearing a disk to be used in a rebuild is strictly for testing purposes. Only disks added to a new slot must be clear so parity will remain valid.

Thanks for the quick reply. What about my question about re-enabling the drive? Is it safe to assume from my log above that my drive is ok and I can "just" re-enable it?

 

Also, any instruction how to preclear a disk in Unraid 6? Can I still do this now given one disk is already disabled?

Yes try rebuild to same disk.

 

Search for preclear plugin, search tips in my sig.

Thanks for your quick and helpful replies. I have started the rebuild process. It indicates that it takes 900 days. Hope this is not a bad sign and will eventually speed up...

And now disk 2 (the faulty one) shows again as "disabled". I did not stop the process, so something went wrong. Attached an error log. Your help is highly appreciated!

tower-diagnostics-20160210-0940.zip

Any thoughts about my log. As described, I removed and readded the faulty disk. Recovery process took around 2-3 minutes (showing 1,000 days remaining) and then the disk got disabled again. Any feedback based on provided log would be appreciated. Thanks!