[Discuss] Backing up the entire filesystem

[Rich Braun]

Rich Pieri notes:
> there is a flaw (or possibly several) in the ext4  [that] causes extent corruption and data loss.
> fsck will *not* repair an ext4 filesystem in this state.

The ext4 filesystem has been the mainstream default on most Linux distros since 2008. I’ve been using a large number of systems using this filesystem, and a lesser number with xfs, ever since. Never once have I had filesystem corruption like with previous filesystems; there are other far-more likely data loss scenarios (like, uh, human error).

I could not find a reference for the “fatal flaw” cited here but here’s an overview of how ext4 stacks up: https://opensource.com/article/18/4/ext4-filesystem.

I’d say if you’re running a mainstream distro with ext4, do your backups, don’t worry about corruption (unless you’re inside the Beltway), and leave the religious flame wars about with filesystem is best to the Linux kernel-contributor experts. If you’re setting up a new system, xfs, btrfs and ext4 are all valid choices.

On the main question of backups: just a couple days ago, Crashplan finally pulled the plug on its Home service. I have two other solutions active for my systems: Duplicati for backup to a local server and to the Backblaze B2 cloud service ($60/year per terabyte), and rsnapshot to another local server. 

I have published a python package ‘secondshot’ to add a few capabilities on top of rsnapshot; see my repos on GitHub / dockerhub / pypi. (github.com/instantlinux).

-rich