​
Overview

A shared filesystem is network-attached file storage that many machines can mount simultaneously. That’s the one thing a volume cannot do — a volume attaches to exactly one machine.
Navigation: Storage → Shared Filesystems. Titled Shared Filesystems, subtitled “Manage shared file storage systems”.

​
When to use one

Shared filesystems are a genuinely different resource with their own lifecycle — they can be started, stopped, and restarted like a machine, which volumes cannot. They also occupy a network, so they have a Network and a Zone.

​
Columns

Expanding a row gives considerably more: Filesystem ID, Name, Provider, Size, Mount Path, State, IP Address, NFS Mount Command (with a copy button), Network, Zone, Created, and Account.
The expanded row’s NFS Mount Command with its copy button is the fastest path to mounting — no need to open the detail page and assemble the command by hand.
Sizes here are in GB directly, not bytes formatted for display. A 500 GB filesystem reads 500 GB, where volumes and snapshots show two decimal places because they store bytes.

​
States

Unlike a volume, a filesystem has a machine-like lifecycle:
Stopping a filesystem breaks it for every machine mounted on it, simultaneously. Unlike detaching a volume from one machine, this is a fleet-wide event. Unmount on the clients first, or expect I/O errors in whatever is using the mount.

​
Filtering

​
Row actions

Destroy and Expunge are different, and the dialogs say so:Destroy Filesystem — “The filesystem will be stopped and moved to a destroyed state. You can recover it later or expunge it permanently.”Expunge Filesystem — “This will permanently delete "" and all its data. This action cannot be undone and the filesystem cannot be recovered.”

​
The detail page

Breadcrumb Storage / Shared Filesystems / {name}, with two sections.

​
Overview

Three copyable rows at the top — Mount command, IP Address, and Export path — then details: State, Provider, File System, Size, Utilization, Zone, Network, Service Offering, Disk Offering, Volume, and Created. Missing values show —.

​
Activity

The Activity section is a placeholder reading “Activity log coming soon” — it is not yet available. For filesystem activity in the meantime, use Events filtered to the resource.

​
Mounting a filesystem

The detail page and the expanded row both give you the mount command. It takes the form {ip}:{export-path}.
Adding a filesystem to /etc/fstab without _netdev can hang a machine at boot while it waits for a mount that networking isn’t up for yet. And a hard mount (the default) to a stopped filesystem makes processes block indefinitely rather than error — soft with a timeout is usually the better choice for anything user-facing.
The machine must be able to reach the filesystem’s Network. A machine on a different network or in another zone won’t reach it, however correct the mount command is.

​
Changing capacity

Change Disk Offering moves the filesystem to a different offering. A Size field appears for customisable offerings. While it runs you’ll see “Changing disk offering in progress…”.
Growing capacity is safe. Shrinking risks data loss if the existing contents exceed the new size — check actual usage (the Utilization field on the detail page) before reducing.

​
Worked example: shared uploads for a web tier

Three web servers must serve the same user-uploaded files.
1

Create the filesystem on the web tier's network

Create Filesystem, sized for growth, on the same network as the web machines. It must be reachable from them.
2

Wait for Running

Created isn’t mountable yet. Start Filesystem if it doesn’t start on its own.
3

Copy the mount command

Expand the row and use the copy button on NFS Mount Command.
4

Mount on every machine, with the same path

Use /mnt/uploads on all three, so your application config is identical everywhere.
5

Add to fstab with _netdev and soft

6

Verify write visibility across machines

Write a file on web-1, read it on web-2. If that works, the shared mount is genuinely shared.
Shared filesystems don’t arbitrate concurrent writes for you. Two machines writing the same file at once can corrupt it. Applications sharing a filesystem need their own locking, or a convention that partitions writes per machine.

​
Empty and error states

Volumes

Single-machine block storage.

Networking

The network clients need to reach.

Events

Activity, while the Activity section is unfinished.