Yes — you mount sdc4 and you unmount everything on the external drive. The rule is: the partition you are reading from must not be mounted; the partition that will hold the image file must be mounted. The path /dev/sdc4/home/alpha/SuRescue cannot exist on any Linux system, because /dev/sdc4 is a device node, not a directory — once the partition is mounted, the file is reached through the mountpoint, for example /mnt/home/alpha/SuRescue.
Two different rules, and why they differ
ddrescue does not read files — it reads raw blocks straight from the device. A mounted filesystem is never passive: the kernel may replay its journal, update metadata, and flush caches to it at any moment. On a healthy disk that is mere noise. On a failing disk, every write is extra stress on already fragile hardware, and any modification to the source while it is being copied means your image is a moving target that can never be consistent.
The destination follows the opposite logic when the target is a file. A file can only exist inside a mounted filesystem, so the partition that will contain it has to be mounted. If you were cloning partition-to-partition or disk-to-disk instead, you would be writing raw blocks over the target, and then the target would also need to be unmounted. In short:
- Source (always): unmounted, whatever the target is.
- Target is an image file: the partition holding the file must be mounted.
- Target is a raw disk or partition: it must be unmounted.
Why /dev/sdc4/home/alpha/SuRescue is not a real path
/dev/sdc4 is how the kernel names the fourth partition of the third disk. It is a special file used to talk to the hardware, not a folder containing your data. The directories /home/alpha live inside the filesystem stored on that partition, and they only become visible somewhere in the directory tree after the partition is mounted.
So if you mount sdc4 at /mnt, your target file is /mnt/home/alpha/SuRescue. In a live session, if you let the desktop environment auto-mount the partition, it usually appears under /media/ — lsblk will show you the exact mountpoint in the MOUNTPOINTS column.
A working sequence for your layout
# identify devices and current mounts
lsblk -f
# make sure nothing on the failing disk is mounted
sudo umount /dev/sda1 /dev/sda5 2>/dev/null
# mount the destination partition
sudo mkdir -p /mnt/target
sudo mount /dev/sdc4 /mnt/target
# confirm there is room for the image
df -h /mnt/target
# rescue sda1 into a file, with a mapfile alongside it
sudo ddrescue -d -r3 /dev/sda1 /mnt/target/home/alpha/SuRescue /mnt/target/home/alpha/SuRescue.map
Here -d asks for direct access to the device and -r3 sets three retry passes; the mapfile lets you interrupt and resume later. Note that sda2 in your listing is just the 1 KiB placeholder for an extended partition — it holds no data and can never be mounted. Also check capacity: your sda1 is about 466 GiB and sdc4 offers 871.7 GiB, so a partition image fits, but a whole-disk image of sda (931.5 GiB) would not fit on sdc4.
Do you actually need the live USB?
The point of a live session is to keep your running operating system away from the drives involved. That is essential when the disk being rescued is the one your system boots from, or when the installed system no longer starts. In your layout neither is true: the patient is the external drive and the internal disk is merely storage for the image, so booting your installed system with its root on sdc3 is a legitimate option.
The trade-offs:
- Installed system: more convenient, but desktop environments tend to auto-mount an external drive as soon as it is plugged in, and background services keep writing to the internal disk. You would have to unmount
sda1andsda5after boot and resist opening that drive in a file manager. - Live USB: nothing of yours runs from or writes to the failing disk, which is the cleanest setup — but you still must mount
sdc4to have somewhere to put the image. Mounting the destination is not "using" the disk under rescue; it is unavoidable and harmless.
Either way, unmounting sdc4 is never the right move in your scenario: without it, ddrescue has no filesystem in which to create SuRescue.
Frequently Asked Questions
Should I mount all of sdc or only sdc4?
Only sdc4. Mounting operates on filesystems, which live in partitions; with a partitioned disk there is no meaningful "mount the whole disk". Mount sdc4 at a mountpoint of your choice and build the target path under it.
Can I boot from my installed system instead of the live USB?
Yes, provided the failing drive is not your system disk and its partitions stay unmounted. With root on sdc3, sdc4 is likely your /home and will be mounted automatically — which is exactly what you want, since the image file lives there.
What if the file manager auto-mounts the external drive by accident?
Unmount it before starting or resuming ddrescue. Be aware that any region copied while the source was mounted and writable may be stale; for a failing drive, the safest response is to restart the image from scratch rather than trust a possibly inconsistent copy.
Where should the mapfile go?
Next to the image, inside the same mounted destination — for example /mnt/target/home/alpha/SuRescue.map. Never place it on the failing drive itself.
How much free space does the destination need?
Plan for the full size of the source partition, not just its used data — ddrescue creates an image of the entire partition. Check with df -h after mounting the destination before you start.
Should I image the whole external disk instead of one partition?
For a failing drive, one whole-disk pass is often preferred because it touches the hardware only once. In your case, however, the 931.5 GiB disk image would not fit on the 871.7 GiB sdc4, so imaging partition-by-partition to separate files is the practical choice.