Charlie Hoffpauir wrote:
On Thu, 25 Apr 2019 17:51:15 -0500, Charlie Hoffpauir
wrote:

I have a couple of Seagate Hard Drives that Hard Disk Semtinel says
are failing. One is model ST2000DM001 and the other ST3000DM008. I've
removed both from my system and recovered the data, but I'm wondering
if there's anything that can be done to "save" the drives, perhaps as
backups, or is that a useless activity and I should just toss them.
Hard Disk Sentinel says one has an expected life of 5 days, and the
other 16 days. The problems reported are bad sectors, for example, for
the 3 TB drive:
56 bad sectors
3288 bad sectors during self test
4267 errors during data transfer
power on time 587 days, 3 hours
Est remaining lifetime 5 days
total start/stop count 11,804

so, can anything be done for them, or or they scrap.

Thanks for all the comments.
I was actually thinking.... I could put one or both in service as one
of my rotating backups. In that manner it would only be used once per
month, for only a few minutes. "5 days" could last months that way.
Yesterday I tried to run Seatools DOS, downloaded from the Seagate
site, but it looks like the GUI version doesn't work (on my setup at
least... it doesn't locate any drives), so I did a "long" format from
within Windows (5+ hours on the 3 TB). Results unchanged.... still
predicting 5 days. I did notice that this drive does run several
degrees hotter than any of my other drives that are in service. The
only one I've seen run hotter is a WD Black 3 TB that runs so hot I'm
afraid to use it in my system. (FYI, SSDs in my system run typically
27-28 C, HDDs about 33-35 C, and this 3TB Seagate pushes 40 C.

Human body temperature is 37C. Another 3C won't hurt.

*******

You're probably thinking of the old days, where one surface was
used solely for servo, and all other surfaces were soft formatted.
When you did a "low level format", N-1 surfaces would receive
new header, new data, write splice, lead-in and so on. The track
was rewritten.

Today, the pattern on each platter is "fixed". There is no "low
level format" possible. You can write the "data" portion of sectors
and that is all.

Formatting a drive will make *no difference* to the
reallocation status of the drive.

---1--- ---2--- ---3--- ---x--- ---5--- -servo- -servo- ---6---
--29---

In that example, sector 29 is being used as a substitute for sector 4.
The servo information is in-line, and disrupts sectors in a sense.

The "format" of the disk is fixed. When you write the disk from end
to end, the servo is never re-written, the sector headers (if there
are any) remain fixed in place for the life of the drive.

Since sector 4 was reallocated, it stays in the reallocation count,
Sector 29 receives the data that sector 4 used to receive. So 29
got a new data section in place of 4 getting it. Sector 29 is probably
on the same track as sector 4, and this is purely diagrammatic detail.

While SCSI has the ability to reset grown defects, ATA drives don't
do that. Only a factory utility of some sort, or a command issued
through the serial interface on the drive, might be able to do that.
It's not intended that a user be able to reset re-allocations on
IDE/SATA drives.

Once the drive runs out of substitutes in a "bad patch", that's when
you might get CRC errors showing up.

What little I've been able to gather, it seems my Seagate drives
only increment "Current Pending", if there is a lack of spares
available. Even on a sick drive with "Reallocations" raw data field
non-zero, the "Current Pending" does not follow what the name
suggests. The implication was that "Current Pending" were sectors
being processed, and a decision made as to whether they were
bad or not. If you write the drive from end to end, the Current
Pending could be decided one way or another, so the count
was supposed to drop to zero. But on Seagate, I don't think
it works that way, for some reason.

Another thing you should know, is "long format" does a
"quick format" (write out a new $MFT) followed by a *read verify*
of the surface. 99.9% of the time spent, is doing reads.

If you want the surface to be *written* you need this
as an example. In an admin command prompt, try

diskpart
list disk
select disk 3 # make *sure* you're on the correct disk!
clean all # 3 hour end-to-end write of every sector
exit

An alternative is to use chrysocome dd in an admin command prompt.
( http://www.chrysocome.net/dd )

dd if=/dev/random of=\\?\Device\Harddisk2\Partition0 bs=1M

The order that

dd --list

lists disks, is the same order as Disk Management, except numbering
starts from zero. So "Disk 3" in Disk Management, might be
"Harddisk2" when using dd.exe . For both of these procedures,
be extra careful you're erasing the correct drive.

The Windows "long format" command on the other hand, is mostly
about reading the surface and updating NTFS $BADCLUS so that
bad clusters are taken out of the cluster pool.

HTH,
Paul