`ddrescue` a directory

README.md

ddrfind - ddrescue on a directory

A non-recursive approach to using ddrescue on a directory.

Chances are, if you have a currupted set of memory, it's not just a few files. It's probably a whole disk. Wanna save it? You can use ddrescue, but that doesn't work on directories, only files. If you want to use it on a whole big directory, you'll need to do each file one by one. That's exactly what this script does.

Installation

$ fish ddrfind.fish

ddrfind.fish

function ddrfind --description 'ddrescue on a directory (using find)'
  if not test -d $argv[1]; or set -q $argv[2]
    echo "Usage: ddrfind <input-dir> <output-dir>"
    exit 1
  end

  set -l dest (pwd)"/"$argv[2]

  mkdir -p $dest

  cd $argv[1]
  for file in (find . -name "*")
    if test -d $file
      # make a new sub-directory
      mkdir -p $dest"/"$file
    else
      # rescue the file
      echo $dest"/"$file
      ddrescue $file $dest"/"$file > /dev/null
    end
  end
  cd -
end

funcsave ddrfind

It can work with both relative and absolute paths. Though better use absolute paths and not work on the dying drive.

The cd in line 11 into the source path is a bummer for bad disks that disconnect every time ddrescue touches a bad block. So I rewrote my fork yesterday to keep the working directory as it is. It does some other nice things too, like preserving metadata, and retry files that differ in size on consecutive runs.

I used this a bunch and I have some improvements:

• Stops on error instead of ignoring it (e.g. file not seekable, mapfile not writable).
• Shows progress (I like watching ddrescue at work).
• Saves mapfiles so if a file has many bad sectors and you don't need it rescued, you can stop, delete the file from inputdir and run it again which will skip already rescued files, because mapfiles exist for each of them.
  • Don't forget to delete the mapfiles when the rescue is done.
• Retries indefinitely. I am watching the process closely so I will intervene if it gets stuck.
• Preserves last modified date and stuff (useful for photos). For keeping owner and permissions see https://askubuntu.com/a/56810/543926.

The changes made:

     else
       # rescue the file
       echo $dest"/"$file
-      ddrescue $file $dest"/"$file > /dev/null
+      ddrescue $file $dest"/"$file $dest"/"$file.mapfile -r -1; or return
+      # preserve last modified date
+      touch --reference=$file $dest"/"$file; or return
     end
   end
   cd -

The altered function:

function ddrfind --description 'ddrescue on a directory (using find)'
  if not test -d $argv[1]; or set -q $argv[2]
    echo "Usage: ddrfind <input-dir> <output-dir>"
    exit 1
  end

  set -l dest (pwd)"/"$argv[2]

  mkdir -p $dest

  cd $argv[1]
  for file in (find . -name "*")
    if test -d $file
      # make a new sub-directory
      mkdir -p $dest"/"$file
    else
      # rescue the file
      echo $dest"/"$file
      ddrescue $file $dest"/"$file $dest"/"$file.mapfile -r -1; or return
      # preserve last modified date
      touch --reference=$file $dest"/"$file; or return
    end
  end
  cd -
end

funcsave ddrfind

Hey. This is exactly what I was looking for but ran into a problem. Never used fish or ddrescue before...
I try to copy a folder from the failing drive to another disk using:
ddrfind '/media/[user]/xyz/! DL Save/2022-12/' /media/[user]/Backups/WD-Updates/

But everytime I try it, the script keeps creating "media/[user]/Backups/WD-Updates/" as new folders inside the source folder (2022-12) instead of on the backup drive and starts copying to the already damaged drive...

I know, I must be missing something pretty simple here, but I just can't figure out how to route the output to the new drive...

I also had these issues. There is something wrong with the paths of this script. I think you had to use the target path relative to the source directory. Like ddrfind /media/failingdrive/important ../../gooddrive/rescued.

Thank you so much, it works! I was trying to get this working for hours now before bothering someone else.