Re: Run ddrescue a second time

[Robert Trevellyan]

The recommended way to proceed is to make a clone of the recovered data (in
this case, the 2TB drive), not the failing drive, and work with that. Then,
if you mess up, you can make another copy of the recovered data instead of
risking another run with the original, failing drive.

Robert Trevellyan


On Mon, Dec 9, 2019 at 3:00 PM Shahrukh Merchant <
address@hidden> wrote:

> On 2019-12-06 6:03 PM, Steve Westmoreland wrote:
> > Hello,
> > I know this email address is for bugs, but I could not find a reference
> in the instructions that I needed.
>
> It is also a de facto forum where more knowledgeable people help out
> others in using ddrescue, as in fact they have helped me in the past. So
> let me try to return the favour.
>
> > I ran ddrescue successfully in my linux box, from my affected 1TB drive
> to a new 2TB drive. I have put the 2TB drive aside for safekeeping and
> later restore (hopefully) of files.
>
> Umm, what do you mean "put aside for safekeeping and later restore"? And
> what do you mean by "successfully"? Did you confirm that all sectors
> were restored? Was the recover quick or did ddrescue have to work hard
> on certain areas for a long time? AFAIK, the ddrescue operation *is* the
> restore operation and *now* is the time to understand how successful the
> restore was. If all sectors were restored with no errors, great. If not,
> you may want to do another "incremental" run of ddrescue with the same
> destination (with possibly different option combinations) to try to
> improve the results, or do secondary processing to figure out which
> files were affected, and so on. But "later" may be too late to try to
> get more data off the "affected" drive (by which I assume you mean
> failing drive).
>
> > I would like to run ddrescue a second time from the affected 1TB drive
> to a new 1TB drive, so that I can look into the new 1TB drive and "poke
> around".
>
> Why (as opposed to poking around, even if only in read-only mode, on the
> already recovered drive)? Each time you run ddrescue on a failing drive
> you're risking more data loss *unless* you're re-running it on an
> existing destination image with its existing mapfile in order to
> *improve* the results on failed sectors.
>
> > When I attempt to run: "ddrescue /dev/sda /dev/sdb /logfile.log", it
> tells me the output file (logfile.log) exists and if I overwrite it,
> existing data will be lost. Understandable.
>
> I'm confused about this part. Ddrescue should use an existing mapfile as
> a starting point to improve a partial rescue. Does it have a way to
> figure out that the mapfile belongs to a different recovery attempt
> (different combination of source and destination)? How does it do that
> (I don't see anything like a drive ID or similar in the mapfile)?
>
> It is perfectly normal to rerun ddrescue multiple times using the SAME
> source, destination and mapfile. What is not normal (and also not likely
> to give any useful outcome) is trying to use one ddrescue job's mapfile
> on a new recovery attempt with a new (virgin) destination drive!
>
> > But can I create a second logfile (mapfile) while maintaining the
> original? Or is this a foolish endeavor?
>
> I don't know about "foolish"--there are legitimate non-obvious reasons
> for doing almost anything if taken in the right context. :-) But it is
> certainly sub-optimal to do a second independent ddrescue run from
> scratch on a failing drive when you already have a presumably successful
> first run. Unless (a) you messed up the first run somehow and don't
> trust it, or (b) you're very confident in the first recovery and are now
> just playing around with ddrescue and want a failing drive for that
> purpose. But given what you said, it seems you have not really examined
> the output of the first ddrescue to see how well it has recovered the
> failing drive, and that would be my priority, especially if additional
> ddrescue "improvement" runs are indicated.
>
> Shahrukh
>
> >
> >
> > Thank you for your attention to this question.
> >
> > Steve
> >
> > [cid:image001.png@01D5AC35.9481DB00]
> > Steve Westmoreland
> > 503-780-0302
> >
> >
> >
>
>