Pause and resume

shell

zb instances pause pg-staging
zb instances resume pg-staging

Behaviour

	Paused
Compute	Pods removed; no compute billed
Storage and backups	Kept and billed as usual; scheduled backups continue from the last state
Connections	Refused with a clear protocol error until you resume (Free-plan instances wake on the next connection, held up to 30 s)
Wake time	10 to 40 seconds depending on engine and size
Engines	All except ClickHouse (pause: unsupported in the catalogue)

Free-tier instances pause automatically after 15 minutes without connections and always wake on connect; after 7 idle days they are archived (pods gone, disk kept) and after a further 90 days deleted with the last backup preserved. Paid instances never pause on their own.

To keep a staging database off overnight, run zb instances pause and zb instances resume from your scheduler (cron, a CI schedule) or call POST /v1/orgs/{org}/instances/{id}/actions/pause and .../actions/resume. Pause and resume are audit events.