Hi, this is my /proc/mdstat


If sda3 is fail, how it's possible that sda2 is ok?
Thankyou

They are different partitions on the physical disk /dev/sda. It is possible that you'd end up with a failed block on one part of the disk that didn't affect other parts.

It would stupid to have two partitions from the same disk as RAID members in the same mirror because the likelihood the entire disk would fail and cause you to lose all RAID members is fairly high. However, in your setup you have two separate RAID sets so it means if the entire disk /dev/sda fails it degrades both RAID sets but doesn't destroy them since the other members are on /dev/sdb.

Were it me and I saw disk errors start happening suddenly I might consider just replacing /dev/sda altogether even if everything hasn't yet failed on the theory that it will get worse over time.