sailfork / t3code-github / .repos/ alchemy-effect/ website/ src/ content/ docs/ gcp/ messaging/ scheduler.mdx

consumeSchedule runs a handler on your Cloud Run service on a cron schedule. Alchemy creates the Cloud Scheduler job and a private path on the service that only the scheduler can call.

Add a schedule

// src/Monitor.ts
import * as GCP from "@alchemy.run/gcp";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/http/HttpServerResponse";

export default class Monitor extends GCP.Function<Monitor>()(
  "Monitor",
  { main: import.meta.url },
  Effect.gen(function* () {
    yield* GCP.CloudScheduler.consumeSchedule(
      "Heartbeat",
      { schedule: "* * * * *", timeZone: "Etc/UTC" },
      (event) => Effect.log(`tick ${event.scheduleTime}`),
    );

    return { fetch: Effect.succeed(HttpServerResponse.text("ok")) };
  }).pipe(Effect.provide(GCP.Run.ScheduleEventSource)),
) {}

schedule is standard cron. event carries the job name and the time the run was scheduled for.

Several schedules on one service

Give each schedule its own name:

yield* GCP.CloudScheduler.consumeSchedule(
  "Heartbeat",
  { schedule: "* * * * *", timeZone: "Etc/UTC" },
  recordHeartbeat,
);

yield* GCP.CloudScheduler.consumeSchedule(
  "DailyRollup",
  { schedule: "0 0 * * *", timeZone: "Etc/UTC" },
  rollUpYesterday,
);

Public service, private schedules

invokerIamDisabled: true opens your fetch routes to everyone. The schedule path still rejects any request without a valid token from Cloud Scheduler.

Use bindings in the handler

Handlers use the same clients as fetch:

const writer = yield* GCP.BigQuery.WriteTable(table);

yield* GCP.CloudScheduler.consumeSchedule(
  "Heartbeat",
  { schedule: "* * * * *", timeZone: "Etc/UTC" },
  (event) =>
    writer
      .insert([{ jobName: event.jobName, receivedAt: new Date() }])
      .pipe(Effect.orDie),
);

Start a job on a schedule

To run a Cloud Run job nightly, point a CloudScheduler.Job straight at it. See Run it on a schedule.

Example

examples/gcp-cron runs two schedules on one service and records each run in BigQuery.

Where next