runtime_jobs
Jobs run a command in a fresh sandbox, once or on a cron schedule. Each run spends the free usage first while it fits the limits without credit, then credit at the sandbox rates, within your spending limits. An unknown outcome is never retried by itself. A run that cannot start says why in blockedReason (credits, capacity, cost_cap, permission, volume) and tries again every 30 seconds. To keep data between runs, attach a volume (compute.volumes). create(input), list, get, pause, resume and cancel(id), trigger(id): run it now, once, beside its schedule (a paused job stays paused; refused with run_in_progress while a run is starting or going), reschedule(id, schedule): a new schedule, the next run at its next time, update(id, input): change command, image, compute, secrets, alert, timeoutSeconds or retry in place (the next run uses them; history stays), runs(id, order?: newest): a job's attempts, oldest first or newest first, run(id): one attempt, logs(id of a run, cursor?): its output.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| id | No | ||
| input | No | create: {name, schedule: {kind: once, at (Unix ms; a past time runs at once)} or {kind: cron, expression (five fields), timezone? (IANA, default UTC)}, compute: {region (us-east), vcpu (1-16, usually 2), cpuMode (shared), cpuFloorMillis (50), memoryMiB (up to 65536, usually 4096), diskMiB (at least 3072), durationSeconds (the paid time per run, at most 3600, at least timeoutSeconds), volumes?: [{volumeId, path (such as /data), mode? (rw, the default: one run at a time, so the next run reads what the last wrote; or snapshot: a read-only copy)}] (at most 4, in the same region as the job)}, command: {argv (no shell: ["bash", "-lc", "..."]), cwd? (under /workspace)}, timeoutSeconds, retry?: {maxAttempts (1-5), backoffSeconds}, secrets?: [{name, secretId}] (from runtime_secrets action jobs_put), maxTotalCostMicros?, image? (an image of the account: id, name, name:tag or name@version; default the standard image), alert?: {afterFailures? (runs in a row, default 1), webhook?: {url (https), secretId? (a jobs secret that signs it; the key must be allowed to reveal it, not only use it)}, email? (true: mail the job's person)}}. update: any of command, image (null: the standard image), compute (the fields given), secrets, alert (null: none), timeoutSeconds, retry, maxTotalCostMicros. | |
| limit | No | ||
| order | No | runs: oldest first (the default) or newest first | |
| action | Yes | ||
| cursor | No | ||
| schedule | No | reschedule: {kind: once, at (Unix ms; a past time runs at once)} or {kind: cron, expression (five fields), timezone? (IANA, default UTC)}. | |
| limitBytes | No | ||
| idempotencyKey | No | Optional: left out, one is made and returned as idempotencyKey. Send it back only to retry this exact call after a temporary error. |