BTRFS is a copy on write filesystem (COW)
Unlike journaling file systems like ext4, BTRFS does not overwrite the data when it is modified, instead it writes the data to a new block and changes metadata to point to the latest version of the file. This provides better protection against power failures.
Snapshots
BTRFS offers fast incremental snapshot creation and tools to backup and restore.
Creating the file system and btrfs partition
parted /dev/nvmexxx
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) mklabel gpt
(parted) mkpart primary btrfs 4MiB 100%
(parted) print
(parted) quit
after the partition is created, create the btrfs file system on it by running
sudo mkfs.btrfs /dev/sdxy
Mounting the partition
Temporary mount with
sudo mount /dev/sdxy /mountpoint
Show info about the btrfs
To see the UUID, size, usage of all btrfs disks.
sudo btrfs filesystem show
See detailed disk usage
sudo btrfs filesystem usage /Subvol
Creating the subvolumes
To create a subvolume called data
sudo btrfs subvolume create /mountpoint/data
List all the subvolumes
To create a subvolume called data
sudo btrfs subvolume list /mountpoint/data
Mounting the subvolumes
to make the mount permanent, edit the ‘/etc/fstab’ file
UUID=DISKUUUD /MOUNTPOINT btrfs defaults,noatime,commit=120,compress=zstd:1 0 0
{: .nolineno file=“/etc/fstab” } These are mount options for the root subvol
- noatime : Do not modify metadata on file read. Reduces disk writes.
- commit=120: Time interval for writing changes to disk. Default is 30. Reduces disk writes.
- compress=zstd:1 : Filesystem level compression. Reduces disk writes.
Create a readonly snapshot of a subvolume
sudo btrfs subvolume snapshot -r /data /data/.snapshots/name
Copying snapshots to other drives using send, receive
Send and receive is faster than rsync. The destination must be also btrfs.
sudo btrfs send /data/.snapshots/name | sudo btrfs receive /mnt/backupDISK/backup
Restoring from a snapshot and make the snapshot writable
Copy the snapshot to the target folder.
sudo btrfs property set -tsf /data/snapshot ro false
Creating backups of your snapshot on a remote machine
btrfs send -p /data/ /data/.snapshots/new_snaphot | ssh root@ip "btrfs receive /mnt/backups"
Cool. Thank you, A10aa.
OpenSuse had recommended btrfs, but I chose ext4 as its more familiar/popular. Also, I was installing Leap in an HDD.
Will try it when I install linux again(athinini ethra kollam aavumoo aavo?)Unrelated, but does anyone know to improve booting time in linux. I’ve looked up the archwiki n followed some of the points there.
Try it. It is really easy to setup backup automation for btrfs with btrbk and Cron https://github.com/digint/btrbk
Boot time optimization
Generally speaking don’t boot from HDD, use faster media if you can. Usually I don’t shutdown my machine 😕 always suspend or hibernate.
Thank you
ZFS >> BTRFS, btw
How? More robust since it came out much earlier? Anything else?
If you haven’t migrated to btrfs already (like me), it may be worth waiting a few more kernel release cycles. There is a good chance that bcachefs will get added to the kernel in the meantime. Bcachefs is everything that btrfs promised to be. It even supports native encryption unlike btrfs (btrfs can only be installed inside a partition encrypted with something like dmcrypt).
Good to know, I mainly use BTRFS for the snapshot feature and super easy RAID0 setups. Bcachefs looks promising, I will give it a try once it is in the kernel and widely tested.
So, I have a 14TB ext hdd formatted as ext4 for a while now. I only have about 700gb left while I type this message. I do notice ext4 being slow sometimes. When the disk was emptier it would reach speeds of 140mb/s while now it often only goes to 50mb/s.
Do you think btrfs would give me better performance?
The disk is mainly used as AI storage or big games.
I think ext4 will give you better perf see this post https://www.phoronix.com/news/Linux-5.14-File-Systems but for big files performance diff is negligible I guess.
I see, thanks!
