btrfs Filesystem Overview

btrfs is a modern copy-on-write (CoW) filesystem designed for Linux that offers several advanced capabilities:

Multi-Device Support: btrfs can span multiple underlying storage devices and supports various RAID configurations with online device addition, removal, and modification.

Copy-on-Write Mechanism: When modifying files, btrfs does not update data in-place. Instead, it copies the data first, makes changes to the copy, then redirects pointers from the old location to the new location. This ensures data integrity and enables efficient snapshots.

Checksum Verification: Both data and metadata are protected with checksums, enabling automatic detection of data corruption.

Subvolumes: btrfs supports logical filesystem subdivisions that can be mounted and managed independently.

Snapshot Support: The filesystem supports creating snapshots, including snapshots of snapshots, enabling efficient backup solutions.

Transparent Compression: Data is automatically compressed during writes and decompressed during reads, improving storage efficiency without application awareness.

Creating a btrfs Filesystem

The mkfs.btrfs command initializes the filesystem. Key options include:

• -L 'LABEL': Sets the filesystem label
• -d <type>: Specifies data RAID profile (raid0, raid1, raid5, raid6, raid10, single)
• -m <profile>: Specifies metadata RAID profile (same options as data)

Viewing Filesystem Information

To display all btrf filesystems on the system:

btrfs filesystem show

Mounting btrfs Filesystems

mount -t btrfs /dev/sdb /mnt/data

Enabling Transparent Compression

Compression can be enabled during mount using the compress option:

mount -o compress=lzo /dev/sdb /mnt/data

Supported algorithms are lzo (faster) and zlib (better compression ratio).

Resizing Filesystem Capacity

To grow a filesystem to maximum available space:

btrfs filesystem resize max /mnt/data

To shrink by a specific amount:

btrfs filesystem resize -10G /mnt/data

Managing Devices in a Filesystem

Adding a new device to an existing filesystem:

btrfs device add /dev/sdd /mnt/data

Removing a device (data migrates to remaining devices):

btrfs device delete /dev/sdd /mnt/data

Data Balance Operations

After adding or removing devices, redistribute data across all devices:

btrfs balance start /mnt/data

Converting RAID Levels

Check current storage allocation:

btrfs filesystem df /mnt/data

Convert data or metadata RAID profiles. For example, converting metadata to RAID5 (requires at least 3 devices):

btrfs balance start -mconvert=raid5 /mnt/data

Convert data profile to RAID1:

btrfs balance start -dconvert=raid1 /mnt/data

Note: Ensure sufficient devices exist for the target RAID level before conversion.

Working with Subvolumes

Subvolumes provide logical partitioning within a btrfs filesystem with out requiring separate partitions.

Create two subvolumes:

btrfs subvolume create /mnt/data/vol_archive
btrfs subvolume create /mnt/data/vol_backups

List existing subvolumes:

btrfs subvolume list /mnt/data

Mount a specific subvolume to a different location:

umount /mnt/data
mount -o subvol=vol_archive /dev/sdb /archive

Verify subvolume isolation by creating a file in one subvolume and confirming it doesn't appear in another:

cp /etc/hosts /archive/
umount /archive
mount /dev/sdb /mnt/data
ls /mnt/data/vol_backups

Delete a subvolume:

btrfs subvolume delete /mnt/data/vol_backups

Creating Snapshots

Snapshots capture the state of a subvolume at creation time. They are writable and space-efficient due to CoW mechanics.

Create a snapshot of an existing subvolume:

btrfs subvolume snapshot /mnt/data/vol_archive /mnt/data/archive_backup

Verify snapshot independence by modifying the source and confirming the snapshot remains unchanged:

cd /mnt/data/vol_archive
touch inventory.log
ls /mnt/data/archive_backup

List all subvolumes and snapshots:

btrfs subvolume list /mnt/data

Practical Example

Create a multi-device btrfs filesystem with RAID1 metadata:

mkfs.btrfs -L production_data -d single -m raid1 /dev/sdb /dev/sdc

Mount with compression enabled:

mount -o compress=zlib /dev/sdb /mnt/prod

Add a third device and rebalance:

btrfs device add /dev/sdd /mnt/prod
btrfs balance start /mnt/prod

Create subvolumes for different departments:

btrfs subvolume create /mnt/prod/engineering
btrfs subvolume create /mnt/prod/marketing

Create weekly snapshots:

btrfs subvolume snapshot /mnt/prod/engineering /mnt/prod/engineering_snap_$(date +%Y%m%d)