Help!!! I just tried to upgrade from 6.6.6 to 6.7rc5 and upon reboot 2 of my 4 cache pool RAID 10 disks are now showing up as unassigned devices (with btrfs formatting). They are both on a marvel controller from the motherboard and the other two are on the Intel controller. I don't know if their drive assignment changed i.e. sd(x) or not. If I stop the array and try to re-add the disks to the pool it warns that it will format them. How do recover? I have included my diagnostics and a screenshot.

 

After further investigation it looks like the disks are still sdi and sdj from the previous boots. I'll include my previous working diagnostic too.

 

tower-diagnostics-20190223-2102.zip

tower-diagnostics-20190219-1610.zip

Edited February 24, 20197 yr by jonnytolz

There are kernel issues with Marvell controllers and v6.7 were devices are not detected, though some are not new.

To fix cache it's safer to unassign all cache devices, start the array so Unraid "forgets" the pool, then stop, reassign all cache devices and start again.

P.S. since you already started the array without the cache mounting a new docker and libvirt images were created on the array, assuming they were on cache before, you should delete them after the pool is working.

Thanks for your help... I used your instructions and the cache pool is now fine. Only my sonar docker was messed up and rather than mess with it, I just set it up over again (only took a few minutes).