Snapshots

Save a server's disk before a risky change, then keep the change or roll it back.

A snapshot is for a change you might want to undo, such as an upgrade or a migration. Taking one is instant. Everything written after it goes into a separate layer, which you then either commit (keep the changes) or revert (throw them away and go back to how the disk was).

Take a snapshot#

On the server’s Snapshots tab, choose Take snapshot. A server can have one snapshot at a time.

You can have it commit itself after a number of hours, from 1 to 720 (30 days). That way a snapshot you forget about doesn’t stay in place.

Commit or revert#

  • Commit merges the changes into the disk and ends the snapshot. Do this once you’re happy with the change.
  • Revert throws away everything written since the snapshot, so the disk is back to how it was.

A snapshot isn’t a backup

It lives on the same disk as the server, so it doesn’t protect against losing the server. Use backups for that.

While a snapshot is in place#

You can’t do these until you commit or revert:

  • rebuild the server
  • take a manual backup
  • restore a backup onto it

From the API#

The snapshot routes need servers:read to list snapshots and servers:write for everything else:

curl -X POST "$API/servers/$SERVER/snapshots" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"autoCommitHours": 24}'

Then POST …/snapshots/{id}/commit or POST …/snapshots/{id}/revert. See the API reference.