[conspire] request verification of plans to partition and format external hard drive
rick at linuxmafia.com
Fri Jul 20 19:59:16 PDT 2007
Quoting Daniel Gimpelevich (daniel at gimpelevich.san-francisco.ca.us):
> IDE drives have now done the exact same thing for many years. When an OS
> detects errors on such a disk, it is because the remappable sectors have
> already all been exhausted, and the drive is already toast.
For whatever reason, the strategy seems to simply _work better_
(reliably) on SCSI drive of my acquaintance, and yet consistently poorly
on ATA ones (of my acquaintance). So, I keep hearing people say "ATA
drives now do the exact same thing" but the results don't reflect that.
I don't have an explanation, but this is why I'm _much_ more likely to
want to run badblocks on an ATA drive of uncertain quality. On a SCSI
one, I just put it on the spare machine's Adaptec HBA, do Ctrl-A to
go into the SCSI BIOS, and let it grind through a low-level format,
which appears to rebuild the hotfix area. Nothing to my acquaintance
does that on ATA.
Yes, you're presumably right about the reason for catastrophic
accelerating failure on an ATA drive that's used up its hotfix table,
and about the desirability of running the smartmon daemon. I wasn't
intending to get into that, as it vastly exceeds the needs of Darlene's
> For the above reason, it's necessarily quite useless in any case. Recent
> distro installers don't even offer the option anymore.
Using a distro installer to prepare filesystems? How quaint! I prefer
to use whatever is my best-tested live CD. Currently, that would be
More information about the conspire