Guides
Instance Lifecycle
Start, stop, pause, hibernate, reboot, delete — what each does and what still bills.
States
create → provision → active → stopped / paused → active → terminatedWait for active before connecting. Provision log streams in dashboard during create — watch it, don't guess.
Operations
| Action | Effect | Billing note |
|---|---|---|
| Start / Resume | Powers on | Hourly billing resumes |
| Stop | Powers off, keeps local disk | Reduced or no compute charge; local disk retained |
| Pause / Hibernate | Frozen in place | Still holds capacity; check cost breakdown |
| Reboot / Reset | Restart, local disk kept | Billed through restart |
| Delete | Permanent | Local disk + data gone. Grab checkpoints first. Persistent storage plans are separate and not deleted with the instance. |
Mutations return 202 Accepted with a task ID when queued — not done. Poll task status in instance view.
Out of money mid-run
Wallet empty → paid actions return 402 INSUFFICIENT_BALANCE with required vs available cents. Instance may stop. Top up, then start/resume. See Wallet and Top-ups.
Before delete
- Copy checkpoints, datasets, project notes off the box.
- Check Snapshots and Backups if you need a restore point.
- Confirm no teammate still uses it (org workspace instances are shared).