Snapshots & backups

Two ways to save a pod, for two different situations:

• A snapshot rewinds the pod you have, in place. Fast. Reach for it before a risky change.
• A backup is an off-node copy that restores as a new pod. Your original stays put. Reach for it for disaster recovery, or to spin up a copy.

Until paid backups start, the slots are the same on both: every pod gets 5 snapshot slots and 5 backup slots. 4 of each are yours to take by hand; the 5th is reserved for the automatic schedule and you cannot take it manually. The concepts page has the side-by-side table if you want it. Snapshots are free, and stay free. Backups are free until paid backups start; from then each new one is billed by its size and the backup limit goes away.

Getting to the controls

Both live on the pod's own page, one tab apart. Every task below starts here.

1. Open Pods in the console, or click Pods in the side panel. You land on the pods table.
2. Click the pod's name in the Name column. The name is the link, not the row.
3. You land on the pod's page, on the Overview tab. Under the pod's name is a strip of tabs: Overview, Metrics, Snapshots, Backups, Networking, Connection Strings.

Snapshots and Backups are the two this guide is about, and they are laid out the same way: a New button on the left of the bar above the table with a running count beside it ("3 snapshots"), an Auto schedule picker on the right of the same bar, and the table of what you already have underneath. Once a start date for paid backups is announced, the Backups tab adds a Keep the last field beside its picker.

Backups also have a page of their own: Backups in the side panel, under Workspace, lists every backup on the account. See Every backup in one list.

Snapshots

Take one

1. On the pod's page, click the Snapshots tab.
2. Click New, at the left of the bar above the table.

There is no form to fill in - we name the snapshot for you, and "Snapshot queued" confirms it. It runs in the background and lands in a few seconds to a minute. The new row shows Type Manual and Status pending, and the tab refreshes itself every few seconds until the status reads Available.

Four manual snapshots per pod. On the fifth, the console refuses in as many words:

Snapshot limit reached. Maximum 4 manual snapshots per pod
(1 slot reserved for automatic snapshots)

Delete one first: click the trash icon at the end of a row, then Delete in the Delete snapshot dialog. Deleting cannot be undone.

Restore one

1. On the pod's page, click the Snapshots tab.
2. Find the snapshot in the table. Restore stays greyed out unless its Status reads Available.
3. Click Restore at the end of its row.
4. Read the Restore snapshot dialog - if the pod is running, it tells you the pod will be shut down first and then stay stopped - and click Restore.

The pod rewinds in place and shows Restoring while it works. It needs to be Running or Stopped first - you don't have to stop it yourself; the restore takes care of that. A pod that is mid-operation is refused with Pod must be running or stopped to restore a snapshot. When it's done, the pod sits stopped on the snapshot's state - start it again when you're ready.

Restoring is destructive

Anything written after the snapshot is gone. Want to keep the current state too? Take a manual snapshot first, then restore.

Automatic snapshots

1. On the pod's page, click the Snapshots tab.
2. Find Auto schedule, on the right of the same bar as New. It reads "Select schedule" until you set one.
3. Pick one:
   • Daily - every day at 3:00 AM UTC.
   • Weekly - every Sunday at 3:00 AM UTC.
   • Monthly - the 1st of the month at 3:00 AM UTC.
   • Disabled - off.

There is no Save button. Picking is the saving, and "Snapshot schedule updated" confirms it.

Automatic snapshots rotate among themselves, and the rows whose Type reads Auto are the ones at risk. How many are kept depends on how much of the pod's allowance your manual snapshots are using: with four manual snapshots there is room for one automatic one, so each run replaces the last, and with none the automatic ones fill all five slots before the oldest rotates out. Your manual snapshots are never touched by the schedule.

Backups

Take one

1. On the pod's page, click the Backups tab.
2. Click New, in the same spot as on the Snapshots tab.

This is a full off-node export, so it takes longer than a snapshot. The row moves through pending and running before it reads Available, and the Size column fills in once it lands. You cannot delete a backup while it is running.

Until paid backups start, it's four manual backups per pod, same as snapshots, and the same refusal at the fifth. Once they start there is no limit, and each new backup is billed from the moment it reads Available until you delete it.

Automatic backups

1. On the pod's page, click the Backups tab.
2. Find Auto schedule, on the right of the same bar as New, and pick Daily, Weekly, Monthly or Disabled - the same options at the same 3:00 AM UTC as on the Snapshots tab.

As with snapshots, picking is the saving, and "Backup schedule updated" confirms it. Until paid backups start, automatic backups rotate within the pod's slots exactly like automatic snapshots.

Once a start date is announced, a Keep the last field sits next to Auto schedule. It says how many automatic backups the pod keeps from the start date: 5 unless you change it, anything from 1 to 365. Type a number and press Enter or click away; "Automatic backups now keep the last 7" confirms it. From the start date, each time a scheduled backup finishes, the pod's automatic backups taken since the start date are trimmed to that number, oldest first. Automatic backups from before the start date don't count; they stay free until their Free until date. Manual backups and backups you have kept are never pruned.

Restore one (into a new pod)

Restoring a backup builds a new pod from the copy and leaves your original alone.

1. On the pod's page, click the Backups tab.
2. Find the backup. Restore stays greyed out unless its Status reads Available.
3. Click Restore at the end of its row.
4. The Restore backup dialog spells out that this creates a NEW pod and does not modify the original. Click Restore.

The console doesn't ask you to name the new pod - we pick a name and the toast tells you which one:

Restore started - new pod "cool-otter" is being created

(The API takes an optional name if you care what it's called.)

You stay on the backups tab; the new pod comes up in Restoring over in Pods. Because it creates a pod, a restore uses a slot from your pod limit and needs at least $3 credit.

A pod's Backups tab only lists that pod's backups. A backup whose pod you have since deleted is on the Backups page instead, and restores from there the same way.

A backup marked On request has Request where Restore would be. It was made before our move to new backup storage, so our support team restores it for you: Request opens a support ticket with the backup's details filled in. These older backups are deleted on the date shown under the badge, announced by email at least 30 days ahead.

Every backup in one list

1. Click Backups in the side panel, under Workspace, or open Backups directly.

You get every backup on the account, newest first, including backups of pods you have deleted. The Pod column names the pod each one came from, as it was called when the backup was taken. Type reads Manual, Auto, or Final for the backup taken before a pod was deleted over an unpaid invoice. Every row has Restore (into a new pod, as above), or Request on older backups marked On request, and a trash icon to delete it. A free backup also has a Keep link.

What backups cost

Backups are free until paid backups start, on a date we announce by email at least 30 days ahead and show on the pricing page. From that date each new backup is billed by the hour for its size, per GB a month (see the pricing page), until you delete it. Deleting a pod does not delete its backups, so clear out the ones you no longer want from the Backups page.

Once a start date is announced, the Size column on the pod's Backups tab and on the Backups page shows each backup's monthly cost under its size, or a Free until badge with a date. The free ones are backups taken before the start date, plus final backups: they stay free until the date on the badge, and are deleted then unless you keep them. The account owner gets an email listing them a week before. Paid backups has the details, including how final backups work.

Keep a free backup

Once a start date is announced, any free backup can be kept: it then stays until you delete it, at the normal price. Backups marked On request are the exception and can't be kept.

1. Open the pod's Backups tab, or the Backups page.
2. Find the backup. Under the Free until badge in its Size column is a Keep link, shown once the backup's Status reads Available.
3. Click Keep. The Keep backup dialog tells you what it will cost a month.
4. Click Keep.

"Backup kept" confirms it, and the badge gives way to the monthly cost. Billing for it starts at the later of the start date and the moment you kept it, so keeping one before the start date costs nothing until then.

Clone a pod

Want a copy of a live pod without dealing with a saved backup first? Clone it - same Region, same network, brand-new pod. It also uses a pod slot and needs credit.

1. On the pod's page, click the ... button at the top right of the header (to the right of Shell, when the pod is running and has one).
2. Choose Clone. It is only in the menu while the pod is Running or Stopped.
3. Confirm in the Clone pod dialog, which warns you that the copy gets its own address and its own billing, by clicking Clone.

You land back on the pods table, where the copy shows up as Cloning until it is Running.

Which do I want?

Snapshot before you upgrade a package, edit a config, or try something you might regret - you want the fastest possible undo on this pod. Back up when you'd want the pod back even if the original were gone - a bad disk day, a fat-fingered rm -rf, or migrating to a fresh copy.

Belt and braces: turn on an auto snapshot and an auto backup on anything holding data you can't recreate. Snapshots cost nothing, so let them carry the frequent checkpoints; once paid backups start, set Keep the last to the number of backups you would actually restore from.

Via the API

Everything here is in the API too - see snapshots and backups.