](https://gagor.pro/2025/04/backup-from-broken-nvme-drive-with-ddrescue/images/cover.webp)
I was preparing to move data from my old PC to the new one. Just wanted to backup one of my NVMe drives but when I tried standard dd, it failed:
root@box:~# dd if=/dev/nvme1n1 of=/storage/backup/disk.img obs=1M
dd: error reading '/dev/nvme1n1': Input/output error
...
It was my second drive, which I do not care that much but still wanted to make a backup so I could restore what’s possible on new system. I started looking for option to force dd to skip problematic bytes/blocks - whatever possible - and that’s how I found ddrescue1. I heard about it, but I didn’t had occasion to use it before.
root@box:~# ddrescue -f -n /dev/nvme1n1 /storage/backup/file.img /storage/backup/file.log
GNU ddrescue 1.27
Press Ctrl-C to interrupt
ipos: 1024 GB, non-trimmed: 233472 B, current rate: 33996 kB/s
opos: 1024 GB, non-scraped: 0 B, average rate: 258 MB/s
non-tried: 92602 kB, bad-sector: 0 B, error rate: 16384 B/s
rescued: 1024 GB, bad areas: 0, run time: 1h 6m 4s
pct rescued: 99.99%, read errors: 9, remaining time: 1s
time since last successful read: 0s
Copying non-tried blocks... Pass 1 (forwards)
ipos: 856031 kB, non-trimmed: 385024 B, current rate: 10264 kB/s
opos: 856031 kB, non-scraped: 0 B, average rate: 258 MB/s
non-tried: 6619 kB, bad-sector: 0 B, error rate: 24576 B/s
rescued: 1024 GB, bad areas: 0, run time: 1h 6m 6s
pct rescued: 99.99%, read errors: 13, remaining time: 1s
time since last successful read: 0s
Copying non-tried blocks... Pass 2 (backwards)
ipos: 917962 kB, non-trimmed: 471040 B, current rate: 60645 kB/s
opos: 917962 kB, non-scraped: 0 B, average rate: 258 MB/s
non-tried: 393216 B, bad-sector: 0 B, error rate: 106 kB/s
rescued: 1024 GB, bad areas: 0, run time: 1h 6m 7s
pct rescued: 99.99%, read errors: 15, remaining time: 1s
time since last successful read: 0s
Copying non-tried blocks... Pass 4 (backwards)
ipos: 1024 GB, non-trimmed: 577536 B, current rate: 60645 kB/s
opos: 1024 GB, non-scraped: 0 B, average rate: 258 MB/s
non-tried: 0 B, bad-sector: 0 B, error rate: 106 kB/s
rescued: 1024 GB, bad areas: 0, run time: 1h 6m 7s
pct rescued: 99.99%, read errors: 19, remaining time: 1s
time since last successful read: 0s
Copying non-tried blocks... Pass 5 (forwards)
ipos: 1024 GB, non-trimmed: 0 B, current rate: 81920 B/s
opos: 1024 GB, non-scraped: 233472 B, average rate: 258 MB/s
non-tried: 0 B, bad-sector: 16384 B, error rate: 5632 B/s
rescued: 1024 GB, bad areas: 32, run time: 1h 6m 8s
pct rescued: 99.99%, read errors: 51, remaining time: 1s
time since last successful read: n/a
Trimming failed blocks... (forwards)
Finished
Now I had image. It wasn’t fully correct because of failed reads, but at least I didn’t need to restart dd multiple times. I was able to fix filesystem with fsck, then I had to reinstall few packages to recreate broken files and that’s it.
Enjoyed?
