A year ago I have bought a few disk drives for NAS (RAID5 array). A half year later I have seen different little amount of Current Pending Sector in SMART parameters of disk drives. Offline self test shows, that sometimes each drive have bad LBAs, and sometimes haven’t .
Testing this drive with LINUX utility “badblocks” ( $ badblocks -t random /dev/sdx) shows, that writing data each time is successful, but reading data after writing results with floating amount (from hundreds to thousands) bad blocks (correctable and uncorrectable) in similar areas of disk, but not exactly coinciding. When test writes data counter of Current Pending Sector goes to zero, and when test reads this counter increases.
What is it? Please, help me.