I’m in a process of forcing myself to become familiar with linux (currently on Manjaro just to see how it is). One of the things that has halted this in the past is that I occasionally break important files in the process of trying something new/unfamiliar. I’m not always knowledgeable enough to know what I broke or how to fix it and so I’m looking for a good way to rollback my system if I break it.
I come from both Windows and Mac environments and what I have in mind is something with the same basic functionality as Time machine. I don’t need an elegant interface, but I do want something that takes periodic imagines and allows me to easily roll back any trashfire I did.
Here’s the thing. If I google these kinds of keywords I get drowned in a sea of software, and I don’t quite have the time to try each one. Since I’m definitely on the well-intentioned-idiot end of linux users my main priority is easy of use.
Maybe there’s an even more elegant way of doing what I described rather than the solution I think I need, but if that’s the case I don’t quite know the words I’m trying to google for.
If you’re familiar with Time Machine, you’ll proabably be interested in Timeshift.
Install via the AUR with the package ‘timeshift’.
Unless you’re using a BTRFS filesystem, you’ll want to use the rsync+hardlink option.
My preference is more toward how zfs and btrfs takes snapshots.
Imagine every day or week or year. or all of them. or whenever you install something. You take a snapshot of your system. or whatever piece you want of it.
You keep the number of snapshots you want, And each snapshop only takes up space if the snapshot can not be built from combining your previous snapshots.
You can mount any snapshot as a read only file system and get out the stuff you want, Or if it is a samba share then
Just chose a date and click view.
And if on another computer i use any way i want to back it up to this system.
Also at the moment @SgtAwesomesauce with friends are trying to make a package for allowing you to boot into an snapshot of your drive, So if you fuck up you merely reboot into a snapshot before you fucked up.
And i’m running a very similar things on btrfs atm, However most people wont recommend you running btrfs on raid5/6.
Zfs is way more mature.
Would also suggest zfs. Lookup the
com.sun:auto-snapshot option for making periodic snapshots. There’s also zfs send/recv where you can do say
zfs send pool/[email protected]_weekly-2019-01-01-21h30 | zfs recv backup/fs followed by something like
zfs send -i pool/[email protected]_weekly-2019-01-01-21h30 pool/[email protected]_weekly-2019-02-01-00h00 | zfs recv backup/fs which will require sending and storing only the changes, you can of course pipe this over ssh etc.
Only thing to note is you MUST have backups with zfs, there are no recovery tools for corruption, as it is expected that you have a backup.