On Friday 2016-12-23 23:49, sdm wrote:
The chances of btrfs failing in RAID 1 are very slim, like near 0, unless a controller issue or problems with the drive(s) themselves. mdadm is software RAID and it's not surprising people have issues with it as software RAID is not production-level RAID. Software RAID (poor man's RAID) just isn't that good. If you want real RAID, then you need a true controller card in a server that has ECC ram, with something like a PERC RAID controller card. That has a true battery backup should the server or workstation lose power, so whatever is in the buffer that didn't get written to disk gets written the next time it's powered up.
another point why I also use mdadm RAID on some production machines is more redundancy: Two HDD controllers, half of the disks on controller 1, the other on the second controller. ECC RAM is present. maybe related to firmware or module bugs, I had one controller failing sometimes (problem did not reaccure after some kernel updates), system did not went unscheduled down. Spare drive was instantly been included into the array. wondering if hot spares are supported by btrfs' raid, quick web search gave no clear result. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org