check or scan _without_ retesting sectors?

"J. P. Gilliver (John)" wrote in
:

How many retries is too many though? I _think_ SpinRite (or whatever it
was) had a settable maximum, though that could be disabled.


Even one. Can be, anyway. Hence read-only, as lightly as possible, writing
to images. Maybe once that's done, if the source survived, some
attempt at averaging might warrant a few re-reads, but I wouldn't to it.
Better to wait, use DRY cooling, no humidity, to cool the source by at least
10°C, then try one clean pass to a second image, then more cooling, a third
image, etc, then average the clones wherever they differ. Tedious, no wonder
data recovery can cost so much..

"J. P. Gilliver (John)" wrote in
:

I notices that in protected mode good data seems to copy MUCH faster
than in DOS. I still don't know enough to know exactly why, but I
suspect that if anything causes slow reading of data, you might as well
do it in DOS so you have safe direct hardware access.

I can't remember what protected mode means, but I vaguely do remember
some copying going faster - I think it was floppy to floppy on the same
drive (remember "insert source disc"/"insert destination disc"?), where
I assumed it had a much bigger buffer in the more advanced mode.


Protected more is 32 bits, basically a virtual machine. Which may just mean
that some stores are in RAM, managed by code also in RAM, which speeds things
up. I think DOS could speed handling of floppies by RAM caches too, letting
the user set sizes that may or may not restrict other uses of RAM.

Not related, but general good advice: Franc Zabkar mentioned the hazards of
tools that repeatedly hammer a disk, especially when data is already hard to
read off it. What is not often noticed by people is that the disk may be hard
to read becaus it has already been hammered to hell in 'normal' use. When I
make large backups I always use a different disk, not just a different
partition. Personally, I find it SCARY, that horrible juddering noise when
copying gigabytes around between partitions on the same drive. No matter how
quiet it is in modern drives, I try to avoid it. I prefer to leave that kind
of motion to galvanometer laser scanners where there's no choice in the
matter.

On Sat, 19 Nov 2011 23:18:22 +0000, "J. P. Gilliver (John)"
put finger to keyboard and composed:

In message , Franc Zabkar
writes:
On Sat, 19 Nov 2011 08:56:28 +0000, "J. P. Gilliver (John)"
put finger to keyboard and composed:

I saw [SpinRite] being used on the drive from a DOS ('286-based, IIRR) machine.
I'm pretty sure it was a lot less than 200M.

That's where I first encountered SpinRite. It was great on those old
stepper motor drives.

A much better approach to data recovery is to clone the drive, sector
by sector, using a utility that understands how to work around bad
media. Two examples are ddrescue and dd_rescue, both freeware.

How do they handle bad sectors though? Just give them a default value?

Zeros, I think. Personally I'd like to see bad sectors filled with a
repeating "BAD" text string, or some other value that the user could
define.

A good standalone DOS utility that can perform a surface scan on your
drive is MHDD. It will identify "slow" blocks, ie those that require
several read retries.

How many retries is too many though? I _think_ SpinRite (or whatever it
was) had a settable maximum, though that could be disabled.

The drive automatically retries difficult sectors, without software
intervention. ISTR that a typical maximum number of retries could be
around 10 or so. Some drives support Error Recovery Control (ERC)
which sets the timeout values for reads and writes. That is, if a
drive is unable to complete a read or write request within the ERC
limits, then it aborts the command and returns a failed status.
Samsung calls this feature CCTL (Command Completion Time Limit), while
WD refers to it as TLER (Time Limited Error Recovery).

ISTR that some enterprise drives (eg SCSI or SAS) allow the maximum
number of retries to be set by software via a specific mode page.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

Franc Zabkar wrote in
:

Personally I'd like to see bad sectors filled with a
repeating "BAD" text string, or some other value that the user could
define.


42 41 44 21 00 BB AA DD BAD!.»ªÝ
?
Best I could do for a clean fit to 512 bytes. Or just 4 for text only,
BAD!...

Using the map of bad sectors from the log file, this could be done.

In message ,
Lostgallifreyan writes:
[]
I also saw a forceful recommendation for this:
http://www.datarescue.com/photorescue/v3/drdd.htm
...but I have no situation that can let me judge how good it is when it
really matters.

Strange, the link in your post was to
"http://www.datarescue.com/photorescue/v3/drdd.htm...but". (Obviously I
figured it out.)

Looks useful; always good that it's free, but even so, good. The log
files seem a good idea. Since it seems to be designed among other things
for memory cards as well as hard drives, I can't see why it wouldn't
work with floppies too.
--
J. P. Gilliver. UMRA: 1960/1985 MB++G.5AL-IS-P--Ch++(p)Ar@T0H+Sh0!:`)DNAf

"Bother," said Pooh, as Windows crashed for the umpteenth time.

"J. P. Gilliver (John)" wrote in
:

Strange, the link in your post was to
"http://www.datarescue.com/photorescue/v3/drdd.htm...but". (Obviously I
figured it out.)


I guess some clients don't treat a newline as space.

Looks useful; always good that it's free, but even so, good. The log
files seem a good idea. Since it seems to be designed among other things
for memory cards as well as hard drives, I can't see why it wouldn't
work with floppies too.


Exactly. I snagged a copy because memory cards are something it really might
impress me on some time. I sometimes have trouble with those and haven't
really thought beyond using another, and a backup placed on it. I use an SD
card for GPS logging, and might lose data that DRDD can save. Being able to
resume a long trawl like ddrescue can do seemes to be another specialty of
DRDD which may be directly inspired by the GNU program for all I know...

Lostgallifreyan wrote in
:

it really might
impress me on some time

Or possibly not. Ò^O
I have three hard disks and a RAM drive in my system and it can't see any of
them.

Lostgallifreyan wrote:
"J. P. Gilliver (John)" wrote in
:

Strange, the link in your post was to
"http://www.datarescue.com/photorescue/v3/drdd.htm...but". (Obviously I
figured it out.)


I guess some clients don't treat a newline as space.

Well, they're not the same! A new line is 0AH (hex), and a space is 20H.
Viva la difference! :-)

"Bill in Co" wrote in
m:

Lostgallifreyan wrote:
"J. P. Gilliver (John)" wrote in
:

Strange, the link in your post was to
"http://www.datarescue.com/photorescue/v3/drdd.htm...but". (Obviously I
figured it out.)


I guess some clients don't treat a newline as space.

Well, they're not the same! A new line is 0AH (hex), and a space is 20H.
Viva la difference! :-)


Well, in this case it ought to be, then the URL would not have been run into
the test starting the folowing line.

If you look at most if not all RegExp (regular expression syntax), you'll see
a few other related hex codes, 09 for tab, and A0 (dec 160), looks like a
space but isn't, which can be profoundly useful or extremely annoying
depending on context. Then there's 0D as well as 0A, hopefully in that order,
and ALWATS together in emails, stuff breaks otherwise. And so it goes.

And because of the rules being complex, RegExp calls the whole lot of it
'whitespace'. You can be very specific about each type with RegExp too, but
sometimes the similarity is all that matters, especially if all that space is
a an undesired thing to ignore.

Pus ca change, pus ces't la meme chose! Not very Francais, exactly, nes't
pas? Do you know, I had that slightly sickening pun in a forum sig for
several years...

Lostgallifreyan wrote:
"Bill in Co" wrote in
m:

Lostgallifreyan wrote:
"J. P. Gilliver (John)" wrote in
:

Strange, the link in your post was to
"http://www.datarescue.com/photorescue/v3/drdd.htm...but". (Obviously I
figured it out.)


I guess some clients don't treat a newline as space.

Well, they're not the same! A new line is 0AH (hex), and a space is
20H.
Viva la difference! :-)


Well, in this case it ought to be, then the URL would not have been run
into
the test starting the folowing line.

If you look at most if not all RegExp (regular expression syntax), you'll
see
a few other related hex codes, 09 for tab, and A0 (dec 160), looks like a
space but isn't, which can be profoundly useful or extremely annoying
depending on context. Then there's 0D as well as 0A, hopefully in that
order,

Indeed. First the CR, then the LF. Not sure how things would work
out if it were reversed, though. On further thought, I don't think too
well; the cursor or print head might actually drop down an additional line,
then (not sure, though).

and ALWATS together in emails, stuff breaks otherwise. And so it goes.

And because of the rules being complex, RegExp calls the whole lot of it
'whitespace'. You can be very specific about each type with RegExp too,
but
sometimes the similarity is all that matters, especially if all that space
is
a an undesired thing to ignore.

Pus ca change, pus ces't la meme chose! Not very Francais, exactly, nes't
pas? Do you know, I had that slightly sickening pun in a forum sig for
several years...

Pus or Plus?
I'm assuming that was to mean "the more things change, the more things
remain the same". Ain't it the truth. :-)