Ever since Manjaro 25.0.0 Zetar, unless one has chosen differently, the default filesystem upon a fresh installation of Manjaro is btrfs.
btrfs is a very modern and robust filesystem with an integrated volume manager. Among its many features are…
-
It uses B-trees (“balanced trees”) for just about everything, which means that it’s very fast.
-
It will auto-detect whether the underlying storage device is an SSD or a HDD, and will automatically enable certain optimizations if the underlying device is an SSD. This includes (among other things) the default activation of the
discard=asyncmount option — i.e. a queued TRIM operation — even if it is not explicitly set in the mount options. -
It is copy-on-write. In other words, whenever a file is changed, the file is not overwritten in its entirety, but instead only the modified blocks are written to the drive. On HDDs, this may lead to fragmentation over time, but
btrfshas anautodefragmount option.btrfswill also attempt to avoid fragmentation in the first place by allocating contiguous blocks for data that is yet to be written to the storage device, and it uses delayed allocation, just in case the data is changed in memory before being written out. -
btrfssupports transparent inline compression viazlib,lzoandzstd, whereby it will recognize the compression algorithm used on already existing files. New files will be compressed with the default algorithm as specified in the mount options. -
btrfssupports subvolumes.btrfssubvolumes largely behave as if they were individual partitions, and can thus be mounted with their own dedicated mount options per subvolume, although some mount options — e.g.space_cache— must always be shared with the parent filesystem itself. -
Unlike with separate partitions or
lvm2logical volumes,btrfssubvolumes can be nested — in other words, you can have subvolumes within subvolumes, within yet other subvolumes, and so on — and all subvolumes share the total amount of available free space on the filesystem.
(Note: Unlike partitions,lvm2logical volumes orzfssubvolumes,btrfssubvolumes are not block devices and cannot be treated that way.) -
btrfssubvolumes can be snapshotted with tools liketimeshiftand/orsnapper, in order to create rollback points to a previous version of your subvolume(s). This process — as well as the restoring — is very fast, because it relies on the copy-on-write principle. -
Subvolumes — and snapshots, which themselves are also subvolumes — can be exported to another storage device via “
btrfs send” and “btrfs receive”. (Note: They need to be read-only snapshots, i.e. created with the-rflag.) -
Snapshots can also be used as seeds to populate a new subvolume in a way similar to how
unionfsworks. -
In combination with
grub-btrfs, it is possible to boot into a (read-only) earlier and working snapshot of your system in the event that you’ve managed to break your system somehow, and to then quickly restore your system from the working snapshot. -
btrfsmaintains checksums on both data and metadata. -
btrfssupports RAID levels 0, 1 and 10, as well as data deduplication and file cloning. -
btrfsallows for multiple physically separate filesystems to be combined into a single logical volume.
However, for those new to btrfs, there are a few things to pay attention to. I will list them below.
-
If you find that the command
locatedoes not work, edit the file/etc/updatedb.confand change the line…
PRUNE_BIND_MOUNTS = "yes"… into…
PRUNE_BIND_MOUNTS = "no"The explanation here lies with how
btrfssubvolumes are mounted, which to the system is very similar to a bind-mount. As such, withPRUNE_BIND_MOUNTSset to “yes”,updatedbwill ignore any data residing onbtrfssubvolumes.
-
If your
/bootdirectory resides on abtrfsfilesystem, then you should comment out — i.e. put a “#” in front of the line — all of the compression-related options in/etc/mkinitcpio.conf, and add the following line below them…COMPRESSION=catAfter this you will need to rebuild your
initramfswith…sudo mkinitcpio -PThe change will persist for future system updates.
The explanation for this lies with the fact that
btrfsalready uses compression of its own, and having two compression algorithms running against the same file may actually yield a file that’s bigger than if it were not compressed at all. Therefore, you do not wantmkinitcpiocompressing your kernel image and yourinitramfswhen you’re already using the compression built intobtrfs.
-
Due to how
btrfsphysically organizes the data on disk, you may at some point find that you’re running out of disk space even if you still have plenty of free space according to the traditional tools — e.g.df.This is however easily remedied by running a “
btrfs balance” operation, which will relocate and spread out the data and/or metadata to physically different storage blocks on your drive. This can be done manually, or automatically in the background — for an automatic balancing, you need to have thebtrfsmaintenancepackage installed, which will run a number of scheduled maintenance tasks on your filesystem (if enabled).
-
In line with the above, the
btrfsmaintenancepackage also allows you to enable a timed “btrfs scrub”, which checks the metadata on your filesystem for consistency and will repair it if necessary. It is recommended to run ascrubonce a month, although once a week is also acceptable.You can do this manually or via the timed execution in
btrfsmaintenance. Both thescrubandbalanceoperations are atomic, interruptible — e.g. because you shut down the computer while it was running, or because there was a power outage — and will always leave the filesystem in a consistent state. If the machine was shut down during the operation, then it will resume upon the next boot-up.
In closing, nothing changes for existing installations on different filesystems — the default up until now having been ext4 — and people new to Manjaro can still opt for ext4 (or another Linux-native filesystem) at installation time if they choose to, by selecting manual partitioning in the calamares installer.
