Daniel Carosone wrote:
> Modern disks are supposed to automatically remap bad sectors when
> an error occurs...

> However, it seems that (some?) IDE disks may not do bad sector
> remapping on write if their write-cache is enabled...

> Does anyone have any knowledge to add? ...

Knowlege is too optimistic for what I have, but here it is FWIW:

I have an older WDC WD450AA which has gone bad on three different
occasions and recovered each time -- but very slowly.

Each time the disk remained bad for weeks or months -- it was
only a secondary disk so I was too lazy to yank it.

I've concluded after three episodes that turning the power off
was the event that triggered the recovery.  Powering the machine
off is something I almost never do, which is why the disk stayed
bad for so long each time, I think.

The third failure even produced a scary "SMART status BAD" error
message from the BIOS when rebooting the machine.  I assumed that
the disk had finally used up all the available sectors for re-
mapping and that was that.

BUT, the next time I powered the machine down overnight the BIOS
reported the SMART status was OK again.  I was amazed.  The disk
is still feeling solid right now, so I'm once again not tossing it.

In fact, this installation of NetBSD-current is running from that
disk.  (But it's just a play machine.)