How to fix messaging/QUOTA_EXCEEDED: Sending limit exceeded for message target in Firebase

Inspect the error metadata so you can confirm you hit the target quota and see the Retry-After hint before retrying:

try {
  await messaging.send(message);
} catch (error) {
  console.error('FCM error code:', error.code);
  console.error('Message:', error.message);
  console.error('Retry-After seconds:', error.errorInfo?.retryAfter);
}

The error details highlight whether the limit is per token, topic, or condition and prevent accidental immediate retries that burn the same quota window.

Split your audience and add a short delay before sending to the next slice. A simple throttle loop that waits after each chunk keeps you inside the broadcast limit:

const chunkSize = 200;
const delayMs = 30_000; // 30 seconds per chunk

for (let i = 0; i < tokens.length; i += chunkSize) {
  const batch = tokens.slice(i, i + chunkSize);
  await Promise.all(batch.map(token => messaging.send({ token, notification })));
  await new Promise(resolve => setTimeout(resolve, delayMs));
}

Adjust the delay to spread the traffic so that any individual token or topic only sees one burst within five minutes.

Use the Firebase console quotas dashboard or Cloud Monitoring to see when you approach thresholds. In Cloud Monitoring you can filter for quota metrics such as serviceruntime.googleapis.com/quota/allocation/usage with the quota_metric label that corresponds to FCM. Create alerts for quota/exceeded errors so you know when sends are blocked before the event fires. The linked guide shows how to find quota_metric labels and build charts or alerting policies that surface usage spikes.

For one-time events that still need more throughput, file a quota request at least one month early with dates and audience size. Firebase explicitly honors such requests for events up to one month long but reverts the increase afterward. If you cannot get an increase, start sending 5+ minutes before the event and throttle the rest throughout the interval so you never hit the hard limit.