Adopt a Node.js background job

Use this pattern for a poller or maintenance loop currently started by setInterval(). The business function stays unchanged; the SDK owns when the loop runs and provides cooperative cancellation.

Before

const { pollInvoices } = require('./invoice-service');

const timer = setInterval(() => {
  pollInvoices().catch(console.error);
}, 30_000);

The timer starts with every application instance, has no central status, and can be difficult to stop while a deployment is draining.

After

const { Synapsys } = require('@synapsys-io/sdk');
const { pollInvoices } = require('./invoice-service');

const synapsys = new Synapsys({ workerName: 'billing-api' });

synapsys.endless('invoice-poller', async (stop) => {
  while (!stop.requested) {
    await pollInvoices({ signal: stop.signal });
    await stop.sleep('30s');
  }
});

synapsys.start();

process.on('SIGTERM', async () => {
  await synapsys.stop();
  server.close();
});

Set SYNAPSYS_CORE_URL, SYNAPSYS_CORE_TOKEN, and optionally SYNAPSYS_WORKER_NAME in the deployment environment.

Things to verify

• Delete the old timer so two loops cannot run together.
• Pass stop.signal to network or database calls that support AbortSignal.
• Use stop.sleep() for delays so Stop does not wait for a timer to expire.
• Keep synchronous CPU-heavy work off the event loop; use worker_threads or a child process when appropriate.
• Exercise your SIGTERM path in the same way your container platform stops the app.

Open the execution logs after a test run. Output from the job should appear there and still reach the application's normal logger or console.