harsh-harlequin•3d ago

(Urgent) Question on POST /call Behavior: Concurrency Limit and Queuing

I am writing to inquire about the behavior of outbound calls using the POST /call endpoint when the concurrency limit is exceeded. Based on a recent email announcement, I had expected that Vapi would handle the queuing of calls when the concurrency limit is reached. However, we are receiving an error (e.g., HTTP 429) for requests made on or after the scheduled release date (Nov 21) when the limit is surpassed. Could you please provide support for investigating the cause of this issue? Demand for the service built using Vapi is rapidly increasing, which has created a business-critical situation.

3 Replies

Vapi Support Bot•3d ago

Suggestion:
As a workaround for the concurrency limit, implement a queue management system using Twilio and Redis. For example, you can manually trigger processing of queued calls with the following endpoint:

// Manually process the call queue
app.post('/process-queue', async (req, res) => {
  try {
    await processQueue();
    res.json({ message: 'Queue processing triggered' });
  } catch (error) {
    res.status(500).json({ error: 'Failed to process queue' });
  }
});

// Manually process the call queue
app.post('/process-queue', async (req, res) => {
  try {
    await processQueue();
    res.json({ message: 'Queue processing triggered' });
  } catch (error) {
    res.status(500).json({ error: 'Failed to process queue' });
  }
});

This lets you control when queued calls are retried as capacity becomes available.
See the Vapi call queue management guide for full implementation details. Source: - Call queue management with Twilio

⤵️ Mention @Vapi Support Bot for follow-ups 🐥

Kyle•18h ago

Hi Yasu, Thanks for reaching out — happy to clarify this for you. At the moment, call queueing is only supported for outbound campaigns. Since your system is initiating individual POST /call requests, Vapi won’t automatically queue them when your concurrency limit is exceeded. In this setup, you’ll need to programmatically check how many active calls are currently running before sending a new one to avoid hitting 429 responses. We recommend attaching both the status-update and end-of-call-report webhooks to your server. By listening to these events, you can track active call counts in real time and safely manage your own queueing logic. Best regards, Kyle

harsh-harlequinOP•13h ago

Understood, thank you. So, does this mean that the release mentioned in the email has not been carried out yet? If you have a planned release schedule or roadmap, I would like to know.

(Urgent) Question on POST /call Behavior: Concurrency Limit and Queuing

Did you find this page helpful?