Disks and local storage

node → Disks is where physical hardware becomes usable storage. It has five entries: the disk list itself, and one each for LVM, LVM-Thin, Directory and ZFS.

Preparing storage here is only half the job. The other half is publishing it under Datacenter → Storage so machines can use it — see Storage management.

Checking disk health

node → Disks lists every disk and partition with its size, model, serial number and health.

Column	What to look at
S.M.A.R.T.	PASSED is what you want. Anything else means replace it.
Wearout	On SSDs, how much of the drive’s write endurance is used. This is the number that predicts replacement.
Usage	What the disk is currently used for
Serial	The only reliable way to identify a physical disk

Show S.M.A.R.T. values opens the full attribute list for the selected disk.

Check Wearout, not just PASSED

An SSD approaching the end of its write endurance still reports PASSED right up until it does not. Wearout is the early warning.

Wipe Disk destroys everything on the selected device

It is one click away from the disk list, with nothing else in the way. Before using it, confirm you have the right device by its Serial — not by its device name. Device names can change between boots; serial numbers cannot.

Initialize Disk with GPT prepares a blank disk so it can be used. It is a prerequisite for the other panels, not a destructive operation on a disk that is already empty.

Choosing what to build

Build	When
ZFS	Machine disks, when you want snapshots, checksums, compression — and when you want replication, which requires it
LVM-Thin	Machine disks with space allocated as it is used. The default on a standard installation.
LVM	Machine disks with space allocated up front, and the layer to put on top of shared block storage
Directory	A plain filesystem for ISO images, backups and templates

If you might want replication, choose ZFS

Replication only works for machines on ZFS storage. On a fresh node it is worth deciding this now — the interface cannot convert an existing LVM-Thin installation later.

ZFS

node → Disks → ZFS lists any pools with their size, free space, fragmentation and health.

Create: ZFS builds a pool from selected disks. You choose a name, the disks, and a RAID level determining how much redundancy you get and how much capacity you keep.

Detail shows the pool’s layout and each device’s state. Health is the column to watch — it is where a failing member shows up.

LVM and LVM-Thin

LVM lists volume groups with how much is assigned and how much is free. Create: Volume Group builds one from a disk.

LVM-Thin lists thin pools. A default installation has one, which backs the local-lvm storage. Create: Thinpool adds another.

The thin-pool list shows two independent sets of figures, and both can run out:

Columns	Meaning
Usage, Size, Used	Data space
Metadata Usage, Metadata Size, Metadata Used	Bookkeeping space

Watch metadata as well as data

A thin pool whose metadata fills up stops accepting writes even with data space free, and it cannot be recovered from the interface. Check both figures, not just the headline usage.

Thin pools can be oversubscribed

Thin provisioning lets you promise machines more space than physically exists. That is useful right up to the moment it is not: a pool that genuinely fills stops accepting writes, and every machine on it stops with it. Watch usage, and leave headroom.

Directory

Create: Directory formats a disk with a filesystem and mounts it, ready to be added as a Directory storage. Use it for a disk dedicated to ISO images, templates or backups.

Making it available to machines

Some Create: dialogs offer to add the storage to the cluster at the same time. Where they do not, or if you cleared that option, finish the job yourself:

1. Open Datacenter → Storage.
2. Add the matching type — ZFS, LVM, LVM-Thin or Directory.
3. Point it at what you created, set Content, set Nodes, and click Add.

A pool or volume group that exists on the node but was never added here will not appear anywhere machines can use it. That is almost always the explanation when new storage “does not show up”.

If something goes wrong

What you see	What to do
A disk is not offered when creating a pool	It is in use, or has no partition table. Check Usage, then Initialize Disk with GPT.
New storage does not appear when creating a machine	It was prepared on the node but never added under Datacenter → Storage.
“no space left” although the pool looks empty	The thin pool’s metadata is full. Check the metadata columns.
A ZFS pool shows a health warning	A member device is failing. Use Detail to identify it, and replace it.
A disk vanished after a reboot	Device names renumber. Identify disks by serial number.

Still stuck? Contact VM2Cloud support.