Schedules

Cron-style recurring jobs, declared up-front and reconciled on start().

const q = await Eddyq.connect(process.env.DATABASE_URL!, {
  schedules: [
    {
      name: 'nightly.cleanup',
      cron: '0 3 * * *',
      timezone: 'America/Los_Angeles',
      payload: { keepDays: 30 },
    },
  ],
})

q.work('nightly.cleanup', async ({ payload }) => {
  await deleteOldRows(payload.keepDays)
})

await q.start()

How it works

Schedules are durable — they live in eddyq_schedules, not in your worker process. On start(), eddyq reconciles your declared schedules: config against the table:

• New schedules are inserted
• Removed schedules are tombstoned (no new runs, but in-flight runs finish)
• Cron expression or payload changes update the row in place

If you haven't passed a schedules: config, the table stays empty — that's expected. registerQueue / q.work() don't populate it; only declared schedules do.

A single elected leader (the MAINTENANCE_ROLE) is responsible for inserting due jobs into the queue. After a deploy or leader handoff, missed runs are caught up automatically.

Skipping missed runs

By default, missed runs (e.g. during a long deploy) are caught up. To skip them:

{
  name: 'send.daily.digest',
  cron: '0 9 * * *',
  catchup: false, // missed runs are dropped
}

One-off scheduled jobs

For a single delayed job, prefer enqueue({ delayMs }) over a schedule.