Skip to content

Make a volume snapshot

A volume snapshot is a point-in-time copy of a volume's contents. Snapshots are cheap, fast, and the right tool for "I'm about to make a risky change to this data" — they let you roll forward by creating a new volume from the snapshot if anything goes wrong.

For listing, deleting, and restoring snapshots across the project, see Manage snapshots.

Prerequisites

  • An existing volume. Snapshots can be taken whether the volume is Available (detached) or In-use (attached to a running instance) — see the per-state notes below.

When to take a snapshot

  • Before resizing or extending with LVM — recovery point if the resize goes wrong.
  • Before an in-place upgrade of the database, package, or kernel running on the volume.
  • As a periodic restore point (script via the OpenStack CLI for scheduled snapshots).
  • Before deleting the volume, if there's any chance you'll want the data back.

Steps

1. Take the snapshot

  1. Open Volumes under the project's menu, then click the volume's name.

    Screenshot needed

    Volume detail page with the Actions panel on the right.

  2. In the Actions panel on the right, click Snapshot.

  3. In the Create snapshot dialog:

    • Snapshot name — pre-filled as <volume-name>-snap-<YYYY-MM-DD>. Keep that pattern or replace with something more meaningful (e.g. pg-data-before-pg16-upgrade). Descriptive names make restores far easier later.
    • Description (optional) — extra context for teammates.

    Screenshot needed

    Create snapshot modal showing the pre-filled name and the description field.

  4. Click Create snapshot.

2. Snapshotting an in-use volume

The Portal will let you snapshot a volume that's currently attached to a running instance. OpenStack's volume service quiesces the snapshot using the underlying storage driver — for most backends this is crash-consistent, not application-consistent.

For application consistency:

  • Databases — issue the engine's snapshot/freeze command first (FLUSH TABLES WITH READ LOCK for MySQL, pg_start_backup for Postgres, etc.) so the snapshot reflects a clean transaction state.
  • Filesystems with caches — issue sync inside the guest before triggering the snapshot.
  • Critical workloads — detach the volume, snapshot from Available state, re-attach. Slowest but guaranteed consistent.

Verification

  • The new snapshot appears in the Snapshots card on the volume's detail page within a few seconds.
  • It also appears on the project's Snapshots page with the source volume shown alongside.
  • Status moves through Creating → Available as the snapshot completes.

Restore from a snapshot

Snapshots aren't restored over the source volume — they're used to create a brand-new volume:

  1. Open Snapshots under the project menu — see Manage snapshots.
  2. Click Create volume from snapshot on the row.
  3. The new-volume form opens with the source pinned, size locked to the snapshot, and a name suggested. Adjust as needed and click Create volume.
  4. Attach the new volume to an instance and verify the data.

Next steps