On Tue, Oct 29, 2019 at 02:34:45AM +0500, Roman Mamedov wrote: > On Mon, 28 Oct 2019 13:27:32 -0700 > Marc MERLIN <marc@xxxxxxxxxxx> wrote: > > > Out of desperation, I ran hdrecover /dev/sdx on all my drives. It reads the > > whole drive block by block, allowing to re-read a block many times to try > > and rescue data from it, or just re-write it with 0's. > > That one again, ran fine, no error. > > It is weird that this succeeds, usually a "pending sector" means it's > unreadable until overwritten. > > One possibility is that your RAID card either sets up a HPA at the end of each > drive to store metadata there, or just presents them as somewhat smaller than > their actual size to the OS. If the pending sectors happen to be in that > walled off area, then no wonder that no OS tools can get to them. You are correct, the raid card does wall off a very small portion of the drive, but I'm pretty sure it's fewer sectors than the number of pending sectors I already have. > See `hdparm -N`, or if possible compare `blockdev --getsize64` with the same > model drives which are not connected via a RAID controller. Not easy to do, but I do know that indeed not the entire drive is visible by the OS. Sigh, I hate those silly raid cards without real passthrough :( Marc -- "A mouse is a device used to point at the xterm you want to type in" - A.S.R. Microsoft is to operating systems .... .... what McDonalds is to gourmet cooking Home page: http://marc.merlins.org/ | PGP 7F55D5F27AAF9D08