I ran a bunch of test calls and then near the end of batch the search knowledge base tool was saying timeout in Vapi but in my logs for the tool we were responding within less than 2 seconds. Is there a rate limit or something on Vapi that was causing this error because the tool is working perfectly now. Here is one example but I have tons, where tool worked and where then after one point tool no longer worked on the vapi end. Callid: 019c1182-a0b5-777c-9199-d2cbba2f2f7f Tool call id: call_bB2WTHTklXLJjaU2RYyXEKxi Vapi Response Error: [ { "name": "search-knowledge-base", "toolCallId": "call_bB2WTHTklXLJjaU2RYyXEKxi", "type": "function", "serverUrl": "....", "error": "Your server rejected

tool-calls

webhook. Error: timeout of 20000ms exceeded" } ] @Vapi @Vapi Support Bot https://cdn.discordapp.com/attachments/1467940849735172138/1467940850959913010/image.png?ex=698235f3&is=6980e473&hm=b314e646b6a00cf27f41fbda7fed118e0c6bd1158b923632e0a49763c4b6d74f&

Suggestion: Check the Vapi dashboard's Call Logs for your affected call IDs to identify any error messages or patterns during the batch run. This will help you confirm if the tool timeouts were due to webhook response issues, network latency, or other integration problems. Tip: Make sure your webhook always returns HTTP 200 and the required JSON structure, as any deviation can cause Vapi to ignore the response and register a timeout—even if your server responded quickly. For step-by-step debugging, see the [Custom tools troubleshooting guide](https://docs.vapi.ai/tools/custom-tools-troubleshooting). Source: - [Custom tools troubleshooting](https://docs.vapi.ai/tools/custom-tools-troubleshooting)

Hi Qper, Thank you for providing the detailed information and call ID. I've escalated this to our technical team for further investigation into possible rate limiting or backend issues on the Vapi side. We'll update you as soon as we have more information. If you have additional examples or logs, please feel free to share them to assist our investigation.

Okay thanks Praveen please let me know soon.

Hi Qper, Recommendations for Customer (Qper): 1. Reduce batch concurrency - Don't run hundreds of simultaneous calls 2. Smaller batches - Process 10-20 concurrent calls with delays between batches 3. Monitor end-to-end timing - The 20s timeout includes Vapi's internal processing, not just server time 4. Stagger test calls - Add delays between batches The tool is working correctly - this was infrastructure saturation from high concurrent load. Reducing batch size should prevent future timeouts.

Oh okay, well that's at least helpful to know it wasn't me but something more around rate limiting. Thanks for the update

Hi Qper, just wrapping this up. It seems like the tool is functioning well now. If you run into any more issues or have questions down the line, feel free to reach out. Take care.