switching to direct disk access on rerun

[Tom Lund]

Hi,

I have a failing 12TB disk and I have been able to image all but about
140GB of it by using the -a option to keep the read rate reasonable.  The
remaining skipped blocks also appear to be readable, but only at a very low
rate (retries with successively lower -a values as well as the reported
read rates indicate ~30 KB/s).  As it will take a few months to recover the
remaining data at this rate, I am hunting for a way to speed up the
process.  I did not use the -d option in any of my prior attempts and I
wonder if this may increase the read rate?  I am slightly afraid to try
this option, however, since I am worried that it may not be compatible with
my current map file.  So a few questions:

 1. Can I add the -d (and -b) options on subsequent reruns?

 2. If so, will it help?

 3. Is there any other way to speed up reading these stubborn blocks?

I am not sure if this is kosher, but I have been editing the map file so
that the current position points to the start of the first non-tried or
non-trimmed block and then using commands like

ddrescue -R -a 50k --cpass=1,2 -n -N -f /dev/sdb /dev/sdc map.out

Without editing the map file, I found that ddrescue would often start at
either pass 3 or 4 even when I used the --cpass option (thereby ignoring
the -a input).  I am sure there is a better way to cause ddrescue to start
over with passes 1 and 2 (I tried -r 2 but to no avail).  Perhaps someone
can enlighten me on this.  Also note that I am not attempting to trim or
scrape, as I will do that at the very end.  Currently I have only ~180KB
marked as being non-trimmed.

-- 
Thomas Lund, Ph.D.
Global Atmospheric Technologies and
Sciences, Inc
3360 Mitchell Ln
Boulder, CO 80301