ZFS, or Zettabyte File System, is a powerful and advanced file system and logical volume manager known for its data integrity, snapshot capabilities, and scalability. It primarily uses two command-line utilities: zpool for managing storage pools (the underlying storage devices) and zfs for managing datasets (file systems or volumes created within pools).
I. ZFS Pool Management (zpool)
ZFS pools are the fundamental storage units, built from one or more virtual devices (vdevs) like disks, mirrors, or RAID-Z configurations.
1.1 Pool Creation ➕
- Basic Pool: Creates a pool from one or more disks.
sudo zpool create <poolname> <device1> [<device2> ...]Example:sudo zpool create mypool /dev/sda /dev/sdb(creates a striped pool) - Mirrored Pool (RAID 1): Creates a pool with redundant copies of data.
sudo zpool create <poolname> mirror <device1> <device2> [...]Example:sudo zpool create datapool mirror /dev/sda /dev/sdb - RAID-Z Pool (RAID 5/6/7 equivalent): Creates a pool with parity for data protection.
raidz(single parity, similar to RAID 5):sudo zpool create <poolname> raidz <device1> <device2> <device3> [...]raidz2(double parity, similar to RAID 6):sudo zpool create <poolname> raidz2 <device1> <device2> <device3> <device4> [...]raidz3(triple parity):sudo zpool create <poolname> raidz3 <device1> <device2> <device3> <device4> <device5> [...]Example:sudo zpool create archivepool raidz2 /dev/sde /dev/sdf /dev/sdg /dev/sdh
- Adding Devices (to expand pool):
sudo zpool add <poolname> <vdev_type> <device1> [<device2> ...]Example:sudo zpool add mypool mirror /dev/sdc /dev/sdd(adds a new mirrored vdev tomypool) - Adding Hot Spare: Adds a disk that ZFS can automatically use if another disk fails.
sudo zpool add <poolname> spare <device>Example:sudo zpool add datapool spare /dev/sdz
1.2 Pool Status & Information 📊
- List all pools:
zpool list - Check pool status (health, errors, vdevs):
zpool status [<poolname>]Example:zpool status -x(shows status for unhealthy pools only) - Show I/O statistics:
zpool iostat [<poolname>] [<interval>]Example:zpool iostat -v mypool 5(shows verbose I/O stats formypoolevery 5 seconds) - Scrub a pool (data integrity check): Initiates a read of all data to detect and correct silent data corruptions.
sudo zpool scrub <poolname>Example:sudo zpool scrub datapool
1.3 Pool Maintenance & Repair 🔧
- Offline a device: Temporarily takes a device offline for maintenance.
sudo zpool offline <poolname> <device> - Online a device: Brings an offline device back online.
sudo zpool online <poolname> <device> - Replace a device: Replaces a failed or soon-to-fail disk with a new one.
sudo zpool replace <poolname> <old_device> <new_device>Example:sudo zpool replace datapool /dev/sda /dev/sdb(replaces/dev/sdawith/dev/sdb) - Detach a device (from a mirror): Removes a device from a mirrored vdev.
sudo zpool detach <poolname> <device>Example:sudo zpool detach datapool /dev/sdc - Clear errors: Clears error counts on devices.
sudo zpool clear <poolname> [<device>] - Destroy a pool: Permanently deletes a pool and all its data. Use with extreme caution.
sudo zpool destroy <poolname>Example:sudo zpool destroy testpool
1.4 Pool Import/Export 🌐
- Export a pool: Disconnects a pool from the current system, making it safe to move disks.
sudo zpool export <poolname> - List pools available for import:
zpool import - Import a pool: Attaches a previously exported pool to the current system.
sudo zpool import <poolname>Example:sudo zpool import -d /dev/disk/by-id oldpool(importsoldpoolsearching in/dev/disk/by-id)
II. ZFS Dataset Management (zfs)
Datasets are file systems or block devices (zvols) created within a ZFS pool. They share the pool’s storage and can inherit properties.
2.1 Dataset Creation & Deletion ➕➖
- Create a filesystem dataset:
sudo zfs create <poolname>/<datasetname>Example:sudo zfs create mypool/data(mounts by default to/mypool/data) - Create a volume (zvol): Creates a block device that can be used like a disk.
sudo zfs create -V <size> <poolname>/<volumename>Example:sudo zfs create -V 10G mypool/swapvol(creates a 10GB volume) - Destroy a dataset:Deletes a dataset and all its contents.
sudo zfs destroy <poolname>/<datasetname>Example:sudo zfs destroy mypool/olddata- To destroy recursively (including child datasets and snapshots):
sudo zfs destroy -r <poolname>/<datasetname>
- To destroy recursively (including child datasets and snapshots):
- Rename a dataset:
sudo zfs rename <old_name> <new_name>Example:sudo zfs rename mypool/data mypool/webserver
2.2 Dataset Properties ⚙️
ZFS datasets have many properties that can be set and inherited.
- List properties:
zfs get all <poolname>/<datasetname>Example:zfs get compression,mountpoint,quota mypool/data - Set a property:
sudo zfs set <property>=<value> <poolname>/<datasetname>Example:sudo zfs set compression=lz4 mypool/data(enables LZ4 compression)sudo zfs set quota=100G mypool/users(sets a 100GB quota)sudo zfs set recordsize=1M mypool/bigfiles(optimizes for large files)sudo zfs set mountpoint=/var/www mypool/webserver(changes mount point)
- Inherit a property: Resets a property to its inherited value from its parent.
sudo zfs inherit <property> <poolname>/<datasetname>Example:sudo zfs inherit compression mypool/data/subdata
2.3 Snapshots & Clones 📸
Snapshots are read-only copies of a dataset at a specific point in time, consuming no additional space initially.
- Create a snapshot:
sudo zfs snapshot <poolname>/<datasetname>@<snapshotname>Example:sudo zfs snapshot mypool/data@yesterday - List snapshots:
zfs list -t snapshotExample:zfs list -r -t snapshot mypool/data(lists snapshots formypool/dataand its children) - Rollback to a snapshot: Reverts a dataset to the state of a snapshot. Any changes made after the snapshot will be lost.
sudo zfs rollback <poolname>/<datasetname>@<snapshotname>- To rollback when newer snapshots exist, you must destroy the newer snapshots first or use the
-r(recursive) option to destroy them automatically:sudo zfs rollback -r mypool/data@yesterday
- To rollback when newer snapshots exist, you must destroy the newer snapshots first or use the
- Destroy a snapshot:
sudo zfs destroy <poolname>/<datasetname>@<snapshotname> - Create a clone: Creates a writable copy of a snapshot. The clone initially shares data with the snapshot.
sudo zfs clone <poolname>/<datasetname>@<snapshotname> <poolname>/<clonename>Example:sudo zfs clone mypool/data@yesterday mypool/data_clone
2.4 Send & Receive (Replication) 📤📥
ZFS send and receive allow for efficient, incremental backups and replication of datasets and snapshots.
- Send a full snapshot to a file:
sudo zfs send <poolname>/<datasetname>@<snapshotname> > <backup_file>Example:sudo zfs send mypool/data@snap1 > /mnt/backup/data_snap1.zfs - Receive a full snapshot from a file:
sudo zfs receive <new_pool>/<new_datasetname> < <backup_file>Example:sudo zfs receive backup_pool/data_restore < /mnt/backup/data_snap1.zfs - Send an incremental snapshot (changes between two snapshots):
sudo zfs send -i <poolname>/<datasetname>@<previous_snapshot> <poolname>/<datasetname>@<current_snapshot> > <incremental_backup_file>Example:sudo zfs send -i mypool/data@snap1 mypool/data@snap2 > /mnt/backup/data_inc_snap1_2.zfs - Receive an incremental snapshot:
sudo zfs receive <receiving_pool>/<receiving_datasetname> < <incremental_backup_file>Example:sudo zfs receive backup_pool/data_restore < /mnt/backup/data_inc_snap1_2.zfs - Send/Receive over network (replication):
sudo zfs send <pool>/<dataset>@<snapshot> | ssh user@remote_host "sudo zfs receive <remote_pool>/<remote_dataset>"Example:sudo zfs send mypool/data@daily_snap | ssh user@backup_server "sudo zfs receive backup_pool/data_replica"
III. Mounting & Sharing 📁
- Mount all ZFS filesystems:
sudo zfs mount -a - Unmount a ZFS filesystem:
sudo zfs umount <poolname>/<datasetname> - Share via NFS:
sudo zfs set sharenfs=on <poolname>/<datasetname> - Share via SMB/Samba:
sudo zfs set sharesmb=on <poolname>/<datasetname>