ZFS commands

Feed

date: 2023-11-30 09:26:54

categories: linux

firstPublishDate: 2021-06-27 12:21:41

The commands on this page are my notes reading

FreeBSD handbook ZFS chapter

,

OpenZFS - Root on ZFS

and

ZFS Debian wiki

.

Read my other post for reasons to use ZFS

ZFS in Debian Bullseye

and

Setting up root on ZFS

for installing Debian 11 (bullseye) with ZFS.

Another resource is

the Arch Wiki about zfs

.

Concepts

• zpool manages pools
• zfs manages datasets

pool - vdevs - disks
  `- datasets

Usage

• fstab is not used.

Options:

man zfs
man zpool

• create a pool

zpool create poolName /dev/da0

• create a mirror pool

zpool create poolName mirror /dev/ada1 /dev/ada2

• create raidz pool

zpool create poolName raidz /dev/da0 /dev/da1 /dev/da2

• create a dataset

zfs create poolName/datasetName
zfs set compression=lz4 snapdir=visible poolName/datasetName
zfs set copies=2 compression=lz4 snapdir=visible poolName/datasetName

• create a child dataset and all parent datasets

zfs create -p poolName/dataset1/dataset2/childName

• mount/unmount dataset

zfs mount poolName/datasetName
zfs unmount poolName/datasetName

• set mount point in tree

(all files have to be closed)

zfs mountpoint=/path poolName/datasetName

• delete dataset

zfs destroy poolName/datasetName
delete snapshot
zfs destroy poolName/datasetName@snapshotName

• delete pool

zpool destory poolName

• create a dataset snapshot

zfs snapshot poolName/datasetName@snapshotName

The snapshots are available in `dataset mount point`/.zfs/snapshot/

• create a recursive pool snapshot

zfs snapshot -r poolName@snapshotName

• rollback to a snapshot

zfs rollback poolName/datasetName@snapshotName

• check health

zpool status poolName
# show error details (file paths, errors in snapshots,...):
zpool status poolName -v
zpool status -x
all pools are healthy

• when a disk is dead in a raidz pool

zpool offline storage da1
zpool replace storage da1
zpool status storage

• data verification

zpool scrub poolName

• adding and removing devices

zpool attach >> adds disks to a vdev
zpool add    >> adds vdevs to a pool

• replace a functionning device

zpool replace poolName /dev/ada1p3 /dev/ada2p3

• history

zpool history
zpool history poolName

show more info, like snapshot events:

zpool history -i

long history:

zpool history -l

• performance monitoring

zpool iostat
zpool iostat -v

• list datasets

zfs list

• list snapshots

zfs list -t snapshot

• list all

zfs list -t all

• list a dataset and snapshots for this dataset

zfs list -rt all poolName/datasetName

• list snapshots for a dataset

zfs list -rt snapshot poolName/datasetName

• get compression ratio for a dataset

zfs get used,compressratio,compression,logicalused poolName/datasetName

• get all options for a dataset

zfs get all poolName/datasetName

• rename dataset

zfs rename poolName/datasetName poolName/newDatasetName

• rename snapshot

zfs rename poolName/datasetName@snapshotName newSnapshotName

• diff snapshot

diff snapshot to current

zfs diff poolName/datasetName@snapshotName

diff 2 snapshots

zfs diff poolName/datasetName@snapshotName poolName/datasetName@snapshotName2

• set visible snapshot directory

get state:

zfs get snapdir poolName/datasetName
zfs set snapdir=visible poolName/datasetName
zfs set snapdir=hidden poolName/datasetName

• clone snapshot to dataset

zfs clone poolName/datasetName@snapshotName poolName/newDatasetName

show original snapshot name:

zfs get origin poolName/newDatasetName

remove link with snapshot

zfs promote poolName/newDatasetName

• send dataset with ssh

Setup:
+ host:
zfs allow -u user send,snapshot zroot
+ remote:
sysctl vfs.usermount=1
echo vfs.usermount=1 >> /etc/sysctl.conf
zfs create zroot/backup
zfs allow -u user create,mount,receive zroot/backup
# or zfs allow -u receiver compression,mountpoint,mount,create,receive rxpool
chown user /zroot/backup
zfs create zroot/backup/compressed
zfs set compression=lz4 snapdir=visible zroot/backup/compressed

+ command:
zfs send -vp zroot/compressed@e | ssh user@172.16.43.235 zfs recv -Fdv zroot/backup
-v verbose
-p send dataset properties

receive:
-F Force a rollback of the file system to the most recent snapshot
-d Use the full sent snapshot path without the pool name to determine the name of the new snapshot
-v verbose

• send incremental backups

zfs send -vi zroot/compressed@oldSnapshot zroot/compressed@newSnapshot | ssh user@172.16.43.235 zfs recv -dvu zroot/backup
# zfs recv -Fdv also works

- send with ssh example
zfs snapshot zroot/compressed@initial
+ remote:
+ zfs create zroot/tank
+ zfs set compression=lz4 snapdir=visible zroot/tank
+ zfs allow -u user compression,mountpoint,mount,create,receive zroot/tank
+ chown user /zroot/tank
zfs send -v zroot/compressed@initial | ssh user@172.16.43.235 zfs recv -Fd zroot/tank
zfs snapshot zroot/compressed@evening
zfs send -vi initial zroot/compressed@evening | ssh user@172.16.43.235 zfs recv -Fd zroot/tank

• create a pool and a dataset in an empty disk partition

zpool create \
    -o cachefile=/etc/zfs/zpool.cache \
    -o ashift=12 \
    -O acltype=posixacl -O snapdir=visible -O compression=lz4 \
    -O dnodesize=auto -O normalization=formD -O relatime=on \
    -O xattr=sa -O mountpoint=/home/data \
    sandisk /dev/disk/by-id/DISK-part1

Importing a disk

I moved a zfs disk from a computer to second one and I want mount the datasets in it. To do that use `zpool import` to list available disks with their ids and then run `zpool import $diskId` and mount the datasets with the `zfs` command.

zpool import
   pool: thedisk
     id: 15071621791720838351
  state: ONLINE
 action: The pool can be imported using its name or numeric identifier.
 config:

        thedisk     ONLINE
          ada0s1d   ONLINE
zpool import 15071621791720838351

hashtags: #zfs #linux

Feed