Stop, start, and move

Stop the hourly charge without losing anything. Start again on whatever card is free. Move to another GPU without ever losing the box in between.

Stopping a pod captures its Environment and Volume, then releases the machine. The hourly charge stops. Start later and both come back on a fresh card.

This works on every provider Aquanode supports, because it is built out of two things every provider can do: capture the disk, then hand the machine back.

Manual stop and start

From the pod, use Stop. From the terminal:

aq stop <name|id>

Bring it back on a fresh machine:

aq start <name|id>

You choose the card it comes back on with the same --gpu, --gpus, --max-price, and --provider flags that aq up takes.

A stopped pod rents no machine, so there is no hourly cost while it sits. You pay only to hold what it saved, at $0.02/GiB per month.

Auto-stop

Turn on Auto-stop and the pod does it for you when you walk away.

Threshold	What happens
30 minutes with no GPU use	You get a warning
60 minutes with no GPU use	The environment and volume are captured, the machine is released, the hourly charge stops

Toggle it per pod in the console, or at launch with Stop it when it goes idle. From the terminal:

aq autostop <name|id> on
aq autostop <name|id> off

Idle means no GPU use, not no SSH session

The timer watches GPU utilization. An open shell doing nothing does not keep the box alive, and a long training run with no one connected does not get stopped.

The card you start onto does not have to be the card you left. It does not have to be the same model, the same provider, or the same country. It has to be a card that fits your Environment. That is the point of the pod being separate from the machine.

Moving to another GPU

Move to another GPU, from a Running pod, is a Stop followed by a Start on the new offer. The machine is released only after both the environment and volume captures are confirmed, so a failed Start on the new GPU leaves the pod Stopped with its data intact, never stuck between two machines.

aq move <name|id> --gpu <model>

Saved while you work

A running pod's Volume saves itself in the background, on top of whatever Stop already guarantees. This means a provider taking the box away mid-session loses at most the interval since the last save tick, not everything since the last Stop.

The pod and the Volume page show this status directly:

Saved <N> s ago while it is working
Last save failed: <reason> if a tick could not complete, shown loudly rather than leaving a stale "Saved" timestamp on screen

A tick can catch a file mid-write

Unlike a Stop, which only happens once nothing is being written, a periodic save tick can land in the middle of a file being written and capture it torn. This is accepted: the save tick exists to bound how much you can lose to a sudden provider kill, not to guarantee every intermediate tick is a clean checkpoint. Stop remains the clean point to restore from when you have one.

See Volumes for the full mechanism, including retention and how Start decides which point to restore.

The boundary worth knowing

Who stopped the box changes what you get back

If you stop it, the environment and volume are captured on the way out, so you can start it anywhere: any provider we support, or hardware you already own.

If the provider kills it, there was no chance to capture anything on the way out. You are recoverable to the last save tick, not to the state the box was in when it died, and that tick may hold a torn file if one was being written at that moment.

This is the reason auto-stop is worth turning on for a box you plan to walk away from, and why a five-minute-or-so save interval, not a manual save button, is what stands between you and a preemptible box being taken back.

Tearing it down for good

aq down <name|id>

That stops the rented machine. A kept or shared Environment version, and any Volume you choose to keep, are billed until you delete them, so if you are done with the pod entirely, clear those too.