I know TheFineArticle is in Linux land but for Windows people with this issue you might look at Sysinternals Disk2vhd.[0]
It can be run from the online OS itself and it can store the resulting vhd on the same disk it is imaging (with space and disk performance constraints).
I find it handy for turning my freshly superceded gaming machine into a VM on my new machine for easy access to files, before doing whatever with my old hardware.
[0] https://learn.microsoft.com/en-us/sysinternals/downloads/dis...
FYI, while these block level methods do have a use case, parallel rsync and other file level tools are far safer and often faster with less additional load on the disk.
Duplicating the OS/FS behavior hits the decidable problem and you just hope for the best with block level, often you won't notice corruption either.
1) The article's use case is explicitly bootable images.
2) No, most of us don't "hope for the best" with imaging, but would like to actually achieve a reasonable level of confidence. If your approach to data integrity is "you probably won't notice corruption", you don't have an approach to data integrity.
I'm also doing bootable images. But the old way with a r/o initrd, that does then mount the rootfs.
The rootfs can be mkfs and rsynced nicely.
That said, the article is awesome and the idea very clever.
But more to do streaming replication that dd catchup.
initrd with native OverlayFS kernel support is very versatile. ;)
Yet the btrfs, CephFS, or ZFS all have snapshot syncing tricks that make state mirrors far more practical and safe to pull off. =3
1) The article stated that using bootable images for backup was a preference. That doesn’t invalidate asking whether that’s an ideal preference
2) Arguing that it might be better to avoid such methods because of possible problems with data integrity isn’t a lack of an approach to data integrity.