projects / pve-docs.git / blame
| Commit | Line | Data |
|---|---|---|
| aa039b0f DM |
1 | Local ZFS Pool Backend |
| 2 | ---------------------- | |
| 3 | ||
| 4 | Storage pool type: `zfspool` | |
| 5 | ||
| 6 | This backend allows you to access local ZFS pools (or ZFS filesystems | |
| 7 | inside such pools). | |
| 8 | ||
| 9 | Configuration | |
| 10 | ~~~~~~~~~~~~~ | |
| 11 | ||
| 12 | The backend supports the common storage properties `content`, `nodes`, | |
| 13 | `disable`, and the following ZFS specific properties: | |
| 14 | ||
| 15 | pool:: | |
| 16 | ||
| 17 | Select the ZFS pool/filesystem. All allocations are done within that | |
| 18 | pool. | |
| 19 | ||
| 20 | blocksize:: | |
| 21 | ||
| 22 | Set ZFS blocksize parameter. | |
| 23 | ||
| 24 | sparse:: | |
| 25 | ||
| 26 | Use ZFS thin-provisioning. A sparse volume is a volume whose | |
| 27 | reservation is not equal to the volume size. | |
| 28 | ||
| 29 | .Configuration Example ('/etc/pve/storage.cfg') | |
| 30 | ---- | |
| 31 | zfspool: vmdata | |
| 32 | pool tank/vmdata | |
| 33 | content rootdir,images | |
| 34 | sparse | |
| 35 | ---- | |
| 36 | ||
| 37 | File naming conventions | |
| 38 | ~~~~~~~~~~~~~~~~~~~~~~~ | |
| 39 | ||
| 40 | The backend uses the following naming scheme for VM images: | |
| 41 | ||
| 42 | vm-<VMID>-<NAME> // normal VM images | |
| 43 | base-<VMID>-<NAME> // template VM image (read-only) | |
| 44 | subvol-<VMID>-<NAME> // subvolumes (ZFS filesystem for containers) | |
| 45 | ||
| 46 | `<VMID>`:: | |
| 47 | ||
| 48 | This specifies the owner VM. | |
| 49 | ||
| 50 | `<NAME>`:: | |
| 51 | ||
| 52 | This scan be an arbitrary name (`ascii`) without white spaces. The | |
| 53 | backend uses `disk[N]` as default, where `[N]` is replaced by an | |
| 54 | integer to make the name unique. | |
| 55 | ||
| 56 | ||
| 57 | Storage Features | |
| 58 | ~~~~~~~~~~~~~~~~ | |
| 59 | ||
| 60 | ZFS is probably the most advanced storage type regarding snapshot and | |
| 61 | cloning. The backend uses ZFS datasets for both VM images (format | |
| 62 | `raw`) and container data (format `subvol`). ZFS properties are | |
| 63 | inherited from the parent dataset, so you can simply set defaults | |
| 64 | on the parent dataset. | |
| 65 | ||
| 66 | .Storage features for backend `zfs` | |
| 67 | [width="100%",cols="m,m,3*d",options="header"] | |
| 68 | |============================================================================== | |
| 69 | |Content types |Image formats |Shared |Snapshots |Clones | |
| 70 | |images rootdir |raw subvol |no |yes |yes | |
| 71 | |============================================================================== | |
| 72 | ||
| 73 | Examples | |
| 74 | ~~~~~~~~ | |
| 75 | ||
| 76 | It is recommended to create and extra ZFS filesystem to store your VM images: | |
| 77 | ||
| 78 | # zfs create tank/vmdata | |
| 79 | ||
| 80 | To enable compression on that newly allocated filesystem: | |
| 81 | ||
| 82 | # zfs set compression=on tank/vmdata | |
| 83 | ||
| 84 | You can get a list of available ZFS filesystems with: | |
| 85 | ||
| 86 | # pvesm zfsscan |