Cron schedules

Cron schedules

bladekit.Cron declares a cron schedule on a function. Pass it as the schedule= argument to @app.function(...); the control plane's cron scheduler then triggers that function automatically according to the expression, without any caller invoking .remote(). Schedules only take effect for a deployed app — running the same code with blade run stores the schedule in the function's metadata but never fires it.

import bladekit

app = bladekit.app("reports")

@app.function(gpu="RTX-A4500", schedule=bladekit.Cron("0 3 * * *"))
def nightly_job():
    ...  # runs every day at 03:00 UTC once deployed

blade deploy reports.py

bladekit.Cron

bladekit.Cron(
    expression: str,
    *,
    overlap_policy: str = "skip",
)

expression str

A 5-field UTC cron expression: minute hour day-of-month month day-of-week, e.g. "0 3 * * *" (daily at 03:00 UTC) or "*/5 * * * *" (every 5 minutes).

overlap_policy strDefault: "skip"

What to do if the previous execution is still running when the next trigger is due. "skip" skips the new trigger; "allow" starts it alongside the running one.

schedule= on @app.function

@app.function(
    ...,
    schedule: Cron | None = None,
)

schedule CronDefault: None

A bladekit.Cron(...) instance for cron-scheduled execution. Only takes effect in a deployed app — blade run records it in the function's metadata but does not trigger it.

On blade deploy, every function with a schedule contributes one manifest entry: its name, cron_expr, overlap_policy, a reference to the function, and its resource spec (gpu_type, gpu_count, cpu_millicores, memory_mb, timeout_seconds), plus priority/namespace when set. The server (handleCreateDeployment) inserts a cron_schedules row per entry — that row, not your local process, is what fires the function going forward.

CLI

Once an app is deployed, use blade cron to inspect and manually trigger its registered schedules.

blade cron list app_01hxyz

blade cron history cron_01hxyz --limit 50

blade cron trigger cron_01hxyz

Related reference