Snapshots

A snapshot is a point-in-time image of a pod, taken on the spot and restored in place. It captures the whole pod - files, databases, config, all of it. Snapshots are the fastest way to save a known-good state before a risky change and roll back if it goes wrong.

The key word is in place: restoring a snapshot rewinds this pod. If you want a copy that comes back as a separate new pod, that's a backup, not a snapshot.

Slots and limits

Each pod has 5 snapshot slots by default (the exact allowance is per account - see Settings -> Account):

• 4 for manual snapshots you take yourself.
• 1 reserved for the auto schedule (if you turn it on).

When all 4 manual slots are full, delete one before taking another. The auto slot manages itself.

Auto schedule

A pod can take snapshots on a schedule, all at 3:00 AM UTC:

Schedule	When
Daily	Every day.
Weekly	Every Sunday.
Monthly	The 1st of the month.
Disabled	No automatic snapshots.

Each scheduled run rotates out the previous auto snapshot, so the auto slot always holds the most recent one. Your manual snapshots are never rotated by the schedule.

Restoring

Restoring rolls the pod back to the snapshot's state, in place. Two conditions:

• The pod must be Running or Stopped. You don't have to stop it first - the restore takes care of that and the pod shows Restoring while it works.
• The snapshot must be completed.

When the restore finishes, the pod stays stopped - start it again when you're ready.

Restoring is destructive

Everything written after the snapshot was taken is lost. If you might want the current state too, take a fresh manual snapshot before you restore.

Snapshot vs backup

Same idea, different blast radius: a snapshot restores in place, a backup restores as a new pod. The full comparison is in Snapshots vs backups. Keeping either around costs nothing - snapshots and backups are free.

See also

• Snapshots & backups guide
• Snapshots API
• Backups