I have a Maxtor 275 Gig outboard drive.
Spinrite has worked for me since 2001.
I most recently used it in 2007 May on an 80GB SATA disk on a Dell
Precision 380.
The best that the Dell or disk vendors diagnostics could do
was identify a few (8 if I remember correctly) bad blocks affecting
two sectors. Dell wouldn't even help me get the disk to mark the
bad blocks and revector things so that the problem wouldn't move to
other files. I couldn't backup or clone the disk in the state that it
was in.
(I missed most of the details what SpinRite did since I had about
given up on things, went away for a few hours, missed the messages,
and couldn't save the run log.)
SpinRite was able to recover some of the blocks. I think it took
about 10 minutes for each actual bad block, plus the normal scan time.
I had used another program to determine the file with the problem, get
the file from a backup, and clone the fixed disk to a warranty
replacement disk from Dell. (Since I had the backup I didn't bother to
check if all or just some of the data was recovered.)
I went to check it for the first time with Spinrite and was told by
Spinrite that it would take over 48 hours to check the drive with it's
read & write check.
I does take a long time even if no problems are found. If the GRC guy
said 48 hours, then that probably is correct. Note that each actual
bad block may take considerably extra time. In fact, the 10 minutes
per bad block may have been due to the drive itself finally deciding
to revector the bad block, rather than Spinrite recovering any of the
data. (Old PATA disks and floppies used to sometimes get blocks
fixed;
I don't know if new disks ever actually get blocks fixed by Spinrite.
Spinrite will rewrite things to stop drift problems.)
The only thing that I thought Spinrite should be able to do but
wasn't was to give up on a block after very limited retries. I
thought that Spinrite should be able to force revectoring early,
but this isn't done because of the philosophy of not loosing any
data. It also might not be possible to force revectoring in a
generic fashion, so I'm not sure if the option could be implemented
easily.
Does this sound right? Or will that time estimate possibly change
drastically downwards as Spinrite gets going more into the drive?
No if no problems are found. I forget if taking a long time for early
bad blocks messes up the estimated completion time for the rest of the
run or if only the recent speed is used in the remaining time
estimate. I.e., if it takes 1 hour for the first 10 blocks I huge
estimate might be given. I don't remember if the time remaining
estimate is gradually reduced or is very shortly corrected after a few
cylinders without errors are scanned.
How may people would lock their comps up for two days just to check a
single drive?
Spinrite can continue an operation, so you can run overnight for a few
nights in a row if that suits you.
Is there another program that will do this same job and faster than
Spinrite?
Clone the disk, run the manufacturer's diagnostics, clone back, then
do the manufacturer's read diagnostic. This will do most of the
testing and rewriting of the data that Spinrite does. It won't
recover data from weak blocks any better than your clone program.
However, make sure your clone program works. Most clone programs
don't clone block by block, so things don't get back where they were,
which may break some software licenses.
If you need block by block clone you need a forensic cloner, for
hundreds of US$ to about one thousand US$.
You can also get a cloner that will do only limited recovery if asked
and will clone end to start, for several hundred US$. I'm not sure
if you can get this program to do block by block cloning (meaning each
block is copied to the same disk location on the new disk; not meaning
that the blocks are copied one by one. The blocks might be copied
one by one as part of the error handling stuff, but that is not
important as far as software licensing is concerned.)
