Storage management

Every disk, ISO image and backup archive lives on a storage. Storages are defined once for the whole cluster under Datacenter → Storage, and each is told which nodes can reach it and what kinds of content it may hold.

There are two separate steps, in two different places, and confusing them is the most common storage problem:

Step	Where	What it does
1. Prepare the hardware	node → Disks	Turns physical disks into a pool, volume group or filesystem — see Disks and local storage
2. Publish it	Datacenter → Storage	Makes that space usable by machines, with content types and node visibility

A disk prepared in step 1 but never published in step 2 will not appear anywhere machines can use it.

What a new installation already has

Storage	Type	Holds	Shared
local	Directory	Backups, ISO images, container templates, imports	No
local-lvm or local-zfs	LVM-Thin or ZFS	Machine disks and containers	No

Which of the two you get depends on the filesystem chosen at installation. Neither is shared, so on a single node this is all you need, and in a cluster you will want to add something the other nodes can reach as well.

Adding a storage

1. Open Datacenter → Storage.
2. Click Add and choose the type.
3. Give it an ID — a short name. It cannot be changed later.
4. Fill in the type-specific fields (below).
5. Set Content to what it may hold.
6. Set Nodes to the nodes that can reach it. Leave empty for all nodes.
7. Leave Enable ticked and click Add.

Content types

Content	Holds
Disk image	Virtual machine disks
ISO image	Operating-system installation media
Container template	Container base images
Backup	Backup archives
Container	Container root filesystems
Snippets	Small configuration files used at guest start-up
Import	Guest images staged for importing

Only tick what a storage is genuinely for. A storage offering every content type shows up in every picker, which makes it easy to put a backup on the same disk as the machine it protects.

Do not back up to the same disk the machine lives on

It survives a mistake, but not a disk failure. Give backups their own storage — ideally on different hardware.

Choosing a type

On the node’s own disks

Type	Use it when
Directory	A plain filesystem path. Simple, holds any content type. Good for ISO images and backups.
LVM-Thin	The default for machine disks. Space is allocated as it is used.
LVM	Machine disks with space allocated up front. Also the way to use shared block storage — see iSCSI below.
ZFS	Machine disks with snapshots, checksums and compression. Required if you want replication.
BTRFS	An alternative filesystem for machine disks.

ZFS is a decision you make at installation

Replication only works for machines on ZFS. If you expect to need replication — and you do if you want high availability without shared storage — choose ZFS when you install, or add ZFS disks later. Converting an existing LVM-Thin installation is not a GUI task.

Thin storage can overcommit

LVM-Thin and ZFS thin provisioning let you promise machines more space than exists. That is useful, but a pool that genuinely fills up stops accepting writes and machines using it stop with it. Watch usage on the node’s Disks → LVM-Thin or ZFS panel.

On other equipment

Type	Use it when
NFS	A file server exports a share. Simple and widely available.
SMB/CIFS	A Windows or NAS file share.
iSCSI	A SAN presents block devices. Usually used as the base for an LVM storage layered on top — that combination is the standard way to give every node the same shared block storage.
CephFS / RBD	A Ceph cluster — either one this system runs itself, or an external one.
ZFS over iSCSI	A ZFS appliance reachable over iSCSI.
VM2Cloud Backup Server	A dedicated backup server. Deduplicated, verifiable, and can be encrypted.
ESXi	Read machines off an existing ESXi host in order to import them.

Fields by type

Everything shares ID, Content, Nodes and Enable. Beyond that:

Type	Also asks for
Directory	Directory (the path), Shared
LVM	Base storage, Volume group, Shared
LVM-Thin	Volume group, Thin Pool
BTRFS	Path
ZFS	ZFS Pool, Thin provision, Block Size
NFS	Server, Export, NFS Version
SMB/CIFS	Server, Share, Username, Password, Domain, Subdirectory
iSCSI	Portal, Target, Use LUNs directly
CephFS	Monitor(s), Username, FS Name, Secret Key
RBD	Pool, Monitor(s), Username, Keyring, Namespace
VM2Cloud Backup Server	Server, Username, Password, Datastore, Namespace, Fingerprint
ESXi	Server, Username, Password, Skip Certificate Verification

Some fields only appear once you tick Advanced in the dialog.

“Shared” is a promise, not a mechanism

Ticking Shared on a Directory or LVM storage tells the cluster that every node sees the same data at that location. It does not make that true. Tick it only when the underlying storage really is shared — otherwise machines will be started on nodes that cannot see their disks.

Point Ceph storages at your own cluster with one tick

The RBD and CephFS dialogs have a checkbox for using the Ceph this system runs itself. Ticking it fills in the connection details for you; leave it clear only for an external Ceph cluster.

Retention

Every storage has a Backup Retention tab setting how many backups to keep on it: Keep Last, Keep Hourly, Keep Daily, Keep Weekly, Keep Monthly, Keep Yearly, and Maximum Protected.

This is the default for the storage. A backup job has its own retention settings which override it for the backups that job writes.

Maximum Protected caps how many archives may be marked protected. Protected archives are never pruned, so without a cap a handful of them can fill a storage.

Browsing a storage

Select a storage in the tree to see what is on it:

Panel	What you can do
Summary	Whether it is enabled and active, what it holds, and how full it is
Backups	Restore, view the configuration a backup captured, add notes, protect an archive, prune, delete
ISO Images	Upload installation media, or fetch it with Download from URL
CT Templates	Upload, download, browse the supplied template list, or pull an image from a registry
Import	Stage guest images from elsewhere and import them
Permissions	Who may use this storage

Download from URL needs full administrator rights

Uploading is enough for most storage roles, but fetching a file directly from a URL requires Administrator. See Users, roles and permissions.

Removing a storage

Remove takes the storage out of the cluster’s configuration. It does not delete the data — but machines whose disks were on it will no longer start. Move or delete the machines first.

If something goes wrong

What you see	What to do
A storage is not offered when creating a machine	Its Content does not include Disk image, or Nodes excludes the node you are creating on.
An ISO you uploaded is not in the list	You uploaded it to a storage that is not visible from the node the machine is on.
A storage shows as inactive	The node cannot reach it. Check the server address, credentials and network path.
“no space left” although the pool looks empty	A thin pool can exhaust its metadata before its data space. Check both figures on node → Disks → LVM-Thin.
A machine will not start after a node failure	Its disk is on storage only the failed node could see. This is what shared storage or replication prevents.

Still stuck? Contact VM2Cloud support.