Snapshots

A snapshot is a point-in-time copy of a volume. Use them for backups before risky changes, for cloning VMs, or as a recovery point you can promote into a new instance.

Types

Type	What it captures	When to use
Volume snapshot	One disk (root or data) at a single instant	Backups; quick recovery points
VM snapshot	All volumes + RAM	Whole-VM rollback (KVM-dependent — confirm the option appears in the console for your instance)

For a clean volume snapshot, stop write activity first — either stop the VM, or sync and quiesce databases. CloudStack captures the disk as-is.

Take a snapshot

Console

1. Storage → Volumes.
2. Click the volume you want to snapshot (e.g. ROOT-<vm-id>).
3. Click Take Snapshot, name it (e.g. backup-pre-upgrade), confirm.

CLI

# Find the root volume of an instance
VM=$(cmk list virtualmachines name=hello-atlas | jq -r '.virtualmachine[0].id')
VOL=$(cmk list volumes virtualmachineid=$VM type=ROOT | jq -r '.volume[0].id')
 
cmk create snapshot volumeid=$VOL name=backup-pre-upgrade
cmk list snapshots filter=id,name,state,created

Scheduled snapshots

You can automate snapshots on an interval with a retention policy:

cmk create snapshotpolicy volumeid=$VOL intervaltype=DAILY \
  schedule="03:00" timezone="UTC" maxsnaps=7

Without a retention policy (maxsnaps), recurring snapshots accumulate forever and your storage bill will grow. Set a sensible cap.

Console equivalent: Storage → Volumes → click volume → Recurring Snapshots → +.

Restore from a snapshot

You don’t “restore in place” — you create a new volume from the snapshot and attach it.

SNAP=<snapshot-id>
NEWVOL=$(cmk create volume snapshotid=$SNAP name=restored-root | jq -r '.volume.id')
 
# Stop the VM, detach the old root, attach the new one
cmk stop virtualmachine id=$VM
cmk detach volume id=$VOL
cmk attach volume id=$NEWVOL virtualmachineid=$VM
cmk start virtualmachine id=$VM

Or, promote the snapshot to a Template and launch a fresh instance from it:

cmk create template snapshotid=$SNAP name=my-template ostypeid=<linux-ostype>

Billing

Snapshots are billed by the storage they consume — incremental, so successive snapshots typically use less than the volume itself. See Operations for the current rate.

Common failures

• Snapshot stuck in Allocating — CloudStack’s snapshot pipeline is async; wait a few minutes. If it persists, retry. If it persists after retry, contact help@runatlas.is.
• Restore can’t attach the new volume — the source VM must be stopped, and only one volume can occupy the root slot at a time. Detach the old root first.