10 pvesm - Proxmox VE Storage Manager
16 include::pvesm.1-synopsis.adoc[]
30 The {pve} storage model is very flexible. Virtual machine images
31 can either be stored on one or several local storages, or on shared
32 storage like NFS or iSCSI (NAS, SAN). There are no limits, and you may
33 configure as many storage pools as you like. You can use all
34 storage technologies available for Debian Linux.
36 One major benefit of storing VMs on shared storage is the ability to
37 live-migrate running machines without any downtime, as all nodes in
38 the cluster have direct access to VM disk images. There is no need to
39 copy VM image data, so live migration is very fast in that case.
41 The storage library (package `libpve-storage-perl`) uses a flexible
42 plugin system to provide a common interface to all storage types. This
43 can be easily adopted to include further storage types in future.
49 There are basically two different classes of storage types:
53 Allows to store large 'raw' images. It is usually not possible to store
54 other files (ISO, backups, ..) on such storage types. Most modern
55 block level storage implementations support snapshots and clones.
56 RADOS, Sheepdog and GlusterFS are distributed systems, replicating storage
57 data to different nodes.
61 They allow access to a full featured (POSIX) file system. They are
62 more flexible, and allows you to store any content type. ZFS is
63 probably the most advanced system, and it has full support for
67 .Available storage types
68 [width="100%",cols="<d,1*m,4*d",options="header"]
69 |===========================================================
70 |Description |PVE type |Level |Shared|Snapshots|Stable
71 |ZFS (local) |zfspool |file |no |yes |yes
72 |Directory |dir |file |no |no^1^ |yes
73 |NFS |nfs |file |yes |no^1^ |yes
74 |GlusterFS |glusterfs |file |yes |no^1^ |yes
75 |LVM |lvm |block |no^2^ |no |yes
76 |LVM-thin |lvmthin |block |no |yes |yes
77 |iSCSI/kernel |iscsi |block |yes |no |yes
78 |iSCSI/libiscsi |iscsidirect |block |yes |no |yes
79 |Ceph/RBD |rbd |block |yes |yes |yes
80 |Sheepdog |sheepdog |block |yes |yes |beta
81 |ZFS over iSCSI |zfs |block |yes |yes |yes
82 |=========================================================
84 ^1^: On file based storages, snapshots are possible with the 'qcow2' format.
86 ^2^: It is possible to use LVM on top of an iSCSI storage. That way
87 you get a `shared` LVM storage.
93 A number of storages, and the Qemu image format `qcow2`, support 'thin
94 provisioning'. With thin provisioning activated, only the blocks that
95 the guest system actually use will be written to the storage.
97 Say for instance you create a VM with a 32GB hard disk, and after
98 installing the guest system OS, the root file system of the VM contains
99 3 GB of data. In that case only 3GB are written to the storage, even
100 if the guest VM sees a 32GB hard drive. In this way thin provisioning
101 allows you to create disk images which are larger than the currently
102 available storage blocks. You can create large disk images for your
103 VMs, and when the need arises, add more disks to your storage without
104 resizing the VMs' file systems.
106 All storage types which have the ``Snapshots'' feature also support thin
109 CAUTION: If a storage runs full, all guests using volumes on that
110 storage receive IO errors. This can cause file system inconsistencies
111 and may corrupt your data. So it is advisable to avoid
112 over-provisioning of your storage resources, or carefully observe
113 free space to avoid such conditions.
116 Storage Configuration
117 ---------------------
119 All {pve} related storage configuration is stored within a single text
120 file at `/etc/pve/storage.cfg`. As this file is within `/etc/pve/`, it
121 gets automatically distributed to all cluster nodes. So all nodes
122 share the same storage configuration.
124 Sharing storage configuration make perfect sense for shared storage,
125 because the same ``shared'' storage is accessible from all nodes. But is
126 also useful for local storage types. In this case such local storage
127 is available on all nodes, but it is physically different and can have
128 totally different content.
134 Each storage pool has a `<type>`, and is uniquely identified by its
135 `<STORAGE_ID>`. A pool configuration looks like this:
144 The `<type>: <STORAGE_ID>` line starts the pool definition, which is then
145 followed by a list of properties. Most properties have values, but some of
146 them come with reasonable default. In that case you can omit the value.
148 To be more specific, take a look at the default storage configuration
149 after installation. It contains one special local storage pool named
150 `local`, which refers to the directory `/var/lib/vz` and is always
151 available. The {pve} installer creates additional storage entries
152 depending on the storage type chosen at installation time.
154 .Default storage configuration (`/etc/pve/storage.cfg`)
158 content iso,vztmpl,backup
160 # default image store on LVM based installation
164 content rootdir,images
166 # default image store on ZFS based installation
170 content images,rootdir
174 Common Storage Properties
175 ~~~~~~~~~~~~~~~~~~~~~~~~~
177 A few storage properties are common among different storage types.
181 List of cluster node names where this storage is
182 usable/accessible. One can use this property to restrict storage
183 access to a limited set of nodes.
187 A storage can support several content types, for example virtual disk
188 images, cdrom iso images, container templates or container root
189 directories. Not all storage types support all content types. One can set
190 this property to select for what this storage is used for.
198 Allow to store container data.
206 Backup files (`vzdump`).
214 Mark storage as shared.
218 You can use this flag to disable the storage completely.
222 Maximum number of backup files per VM. Use `0` for unlimited.
226 Default image format (`raw|qcow2|vmdk`)
229 WARNING: It is not advisable to use the same storage pool on different
230 {pve} clusters. Some storage operation need exclusive access to the
231 storage, so proper locking is required. While this is implemented
232 within a cluster, it does not work between different clusters.
238 We use a special notation to address storage data. When you allocate
239 data from a storage pool, it returns such a volume identifier. A volume
240 is identified by the `<STORAGE_ID>`, followed by a storage type
241 dependent volume name, separated by colon. A valid `<VOLUME_ID>` looks
244 local:230/example-image.raw
246 local:iso/debian-501-amd64-netinst.iso
248 local:vztmpl/debian-5.0-joomla_1.5.9-1_i386.tar.gz
250 iscsi-storage:0.0.2.scsi-14f504e46494c4500494b5042546d2d646744372d31616d61
252 To get the file system path for a `<VOLUME_ID>` use:
254 pvesm path <VOLUME_ID>
260 There exists an ownership relation for `image` type volumes. Each such
261 volume is owned by a VM or Container. For example volume
262 `local:230/example-image.raw` is owned by VM 230. Most storage
263 backends encodes this ownership information into the volume name.
265 When you remove a VM or Container, the system also removes all
266 associated volumes which are owned by that VM or Container.
269 Using the Command Line Interface
270 --------------------------------
272 It is recommended to familiarize yourself with the concept behind storage
273 pools and volume identifiers, but in real life, you are not forced to do any
274 of those low level operations on the command line. Normally,
275 allocation and removal of volumes is done by the VM and Container
278 Nevertheless, there is a command line tool called `pvesm` (``{pve}
279 Storage Manager''), which is able to perform common storage management
288 pvesm add <TYPE> <STORAGE_ID> <OPTIONS>
289 pvesm add dir <STORAGE_ID> --path <PATH>
290 pvesm add nfs <STORAGE_ID> --path <PATH> --server <SERVER> --export <EXPORT>
291 pvesm add lvm <STORAGE_ID> --vgname <VGNAME>
292 pvesm add iscsi <STORAGE_ID> --portal <HOST[:PORT]> --target <TARGET>
294 Disable storage pools
296 pvesm set <STORAGE_ID> --disable 1
300 pvesm set <STORAGE_ID> --disable 0
302 Change/set storage options
304 pvesm set <STORAGE_ID> <OPTIONS>
305 pvesm set <STORAGE_ID> --shared 1
306 pvesm set local --format qcow2
307 pvesm set <STORAGE_ID> --content iso
309 Remove storage pools. This does not delete any data, and does not
310 disconnect or unmount anything. It just removes the storage
313 pvesm remove <STORAGE_ID>
317 pvesm alloc <STORAGE_ID> <VMID> <name> <size> [--format <raw|qcow2>]
319 Allocate a 4G volume in local storage. The name is auto-generated if
320 you pass an empty string as `<name>`
322 pvesm alloc local <VMID> '' 4G
326 pvesm free <VOLUME_ID>
328 WARNING: This really destroys all volume data.
334 List storage contents
336 pvesm list <STORAGE_ID> [--vmid <VMID>]
338 List volumes allocated by VMID
340 pvesm list <STORAGE_ID> --vmid <VMID>
344 pvesm list <STORAGE_ID> --iso
346 List container templates
348 pvesm list <STORAGE_ID> --vztmpl
350 Show file system path for a volume
352 pvesm path <VOLUME_ID>
359 * link:/wiki/Storage:_Directory[Storage: Directory]
361 * link:/wiki/Storage:_GlusterFS[Storage: GlusterFS]
363 * link:/wiki/Storage:_User_Mode_iSCSI[Storage: User Mode iSCSI]
365 * link:/wiki/Storage:_iSCSI[Storage: iSCSI]
367 * link:/wiki/Storage:_LVM[Storage: LVM]
369 * link:/wiki/Storage:_LVM_Thin[Storage: LVM Thin]
371 * link:/wiki/Storage:_NFS[Storage: NFS]
373 * link:/wiki/Storage:_RBD[Storage: RBD]
375 * link:/wiki/Storage:_ZFS[Storage: ZFS]
377 * link:/wiki/Storage:_ZFS_over_iSCSI[Storage: ZFS over iSCSI]
383 // backend documentation
385 include::pve-storage-dir.adoc[]
387 include::pve-storage-nfs.adoc[]
389 include::pve-storage-glusterfs.adoc[]
391 include::pve-storage-zfspool.adoc[]
393 include::pve-storage-lvm.adoc[]
395 include::pve-storage-lvmthin.adoc[]
397 include::pve-storage-iscsi.adoc[]
399 include::pve-storage-iscsidirect.adoc[]
401 include::pve-storage-rbd.adoc[]
406 include::pve-copyright.adoc[]