In my old setup, I had an Areca card with 2 4tb WD Red's set up as the parity drive for several Seagate 8tb Archive drives.  Everything worked fine, as the array was larger than the Seagates.  In the new setup, I'm using LSI cards and tried to use the same configuration (2 4tb Reds in a Raid 0 array).  I assigned that array as the parity drive and started the rebuild.  It was scheduled to finish sometime in the early hours this morning, but when I got up and checked the system, it said that the parity rebuild failed with 145 errors.  I checked the logs, and it showed that it had attempted to read past the end of the disk.  I checked the size of the array and it was 7,812,499,404 KB according to Unraid.  I checked one of the 8tb drives, 7,814,026,532 KB.  I thought Unraid had the logic to not allow me to assign a parity drive that was too small to cover the largest drive in the system?  If it's supposed to do that, then something didn't work as intended.  I've assigned one of my spare 8tb Archive drives a parity until I can get in an Ironwolf 7200 RPM to use in it's place.  I've also broken the raid array with the 2 4tb's, so I can run preclears on both just to be sure there are no problems with the individual drives.  I seem to recall (NOW anyway, lol), that the LSI cards use some of the space of the drives for their housekeeping on the array, where the Areca cards didn't (or used less), so that would explain the size, but if Unraid is supposed to detect that a drive is too small to cover parity, I think there may be a bug in that logic.

 

I've attached the diagnostic files just in case it needs to be looked at.

media01-diagnostics-20170611-1219.zip

24 minutes ago, heffe2001 said:

seem to recall (NOW anyway, lol), that the LSI cards use some of the space of the drives for their housekeeping on the array, where the Areca cards didn't (or used less),

i think, it's configurable for some cards.. look at your LSI card documentation, there must be an option. i have a Dell 710P RAID card, and there is such an option, but from command-line only..

Does indeed seem like a bug. Close but not close enough.

 

Areca has some setting about truncating a volume to the nearst something. And that if you don't disable that, combining smaller disks to make a larger parity doesn't work as you wind up with something a tad small like you are seeing.

 

Did you check if the LSI card has a similar setting. Doesn't fix the bug but might let you get back to where you were with your configuration.

 

I too use 2 4T drives make 8T parity on an Areca card - have for years. I like this setup! Allows you to buy a new larger drive (just one of them), and be able to use some of your leftover drives to make parity big enough to use it. Just uses an extra drive slot - which can be precious and therefore this is not for everyone.

I looked in the settings, but didn't see anything.  I'd still be using my Areca card, except that it won't fit in the machine I'm using now.  Maximum length/size I can use now is 6.6" long, 1/2 Height.  Not too terribly many Areca cards fit in that space, and I've not found one that does that you can buy reasonably.  Could use a 7" full height, but I'd need at least 3-4 external ports.  I'm currently using 2 LSI boards, a 9201-16E in the full height, a 9207-4i4e in the half height.

 

Yes there's a bug in this case:

1. You define an array which includes parity.
2. You start array which also starts parity sync.
3. You stop before the sync finishes (parity left in "invalid" state, ie, orange dot)
4. You replace the parity device with another device

In step 4 the code does not check if the new parity device is large enough.