dd rescue to same model drive -will it fit

Question

First time using ddrescue. 3Tb drive was showing errors and going offline (in windows only) so I've booted from a stick into systemrescuecd and used

ddrescue -f /dev/sde /dev/sdb /mnt/rescue/mapfile

(the mapfile is on a 3rd drive).

I may have made a mistake here but wanted to get anything copying from the drive that I could ASAP. The target drive is the same make/ model/ size and the only possible one I could use at that moment.

I'm about 40% into the copy with no errors so far but it's hit the slow stage; a reasonable 2 days (with spikes into 100+ but I'm an optimist on that).

i/opos - 959,803MB
Rescued - 959,768MB
non tried - 2040 GB
non-trimmed - 7288kB
non-scraped, errsize, errors - all 0

I have a larger drive coming, hopefully tomorrow (6Tb).

Fully aware that this was risky but wanted to do something before it didn't ever turn back on. It's not making any noises, yet.

In hindsight I may have been better copying to an image (?)

My concern, having time to read more is whether this is likely to complete, or run out of space - I've seen recommendations of copying to a drive 50% larger.

I can see why there may be a small difference in sizes or the drive I'm copying to may have bad sectors itself etc, but not sure why it should be so much larger, if the mapfile is somewhere else, so thought I'd ask in case I've misunderstood something.

Would you

1. Keep the ddrescue running whilst it's working to the other 3tb drive - I guess I can always take an image of the target drive for whatever's copied at least.

2. Stop it to not wear out the source drive and create an new image on the new drive

3. Do something more sensible and I'm an idiot and have misunderstood what I'm doing here.

4. If it does run out of space at the end - what's the process? ddrescue the hopefully good drive to an image, get the last part off this source one and combine? Or am I screwed. I can live with some data loss.

Thanks hugely for any help, sorry for length, hope it's clear.

---

Backups- nope! This one stared failing whilst trying to backup from it, after that failed completely (several powercuts in a short space of time)

Edit: 2 more things I perhaps should mention

1. This is a data drive, rather than boot
2. For the short time I could see it in windows it showed the folder structure, so I'm hoping the partition information will copy ok. Although appreciate I may have some problems with that if the end size is too large (?)

Answer 1

If target is same or larger it will work. You can check with:

blockdev --getsize64 /dev/source /dev/target

If it's smaller it will run out of space, but whether that matters depends how many bytes missing / and what's actually there.

In hindsight I may have been better copying to an image (?)

Image files have downsides too. Plus it needs more space... you can copy a 3TB disk to another 3TB disk but not to an image file on that 3TB disk since a good chunk is lost to partitioning, filesystem metadata, ...

1) Keep the ddrescue running whilst it's working to the other 3tb drive

Yes, although the logfile/mapfile allows you to cancel out anytime and restart with different options. You mentioned it got stuck on a slow area, you can skip ahead or perhaps experiment with the --min-read-rate option. But as long as it's not stuck completely, it might be better to just keep it running for now.

4) If it does run out of space at the end - what's the process?

Depends on whatever is actually on the drive and how much is missing.

Sometimes it doesn't matter at all, for example LVM often has some unusable data at the end (depends on physical extent size and offset) and most filesystems care a lot more about the start of the drive rather than its end.

If there is a small chunk missing you could use dmsetup linear mapping to create a new device with additional space, without having to copy everything all over again. But most likely it's not a serious problem in the first place.