Error: ETIMEDOUT

ETIMEDOUT in cloud environments differs from local behavior in ways that confuse developers who only test locally. Local HTTP connections to external APIs succeed in under 50ms on a fast fiber connection. The same code in a Lambda function, App Engine instance, or Kubernetes pod may time out due to NAT gateway latency, VPC routing overhead, or egress security rules filtering the connection. AWS Lambda has a hard execution timeout (default 3 seconds, max 15 minutes). If your function makes an HTTP request to a slow API and the request exceeds the Lambda timeout, you see ETIMEDOUT with an abrupt process termination rather than the HTTP client's own timeout. This looks like the HTTP client is at fault, but the cause is the Lambda timeout ceiling. Connection pooling is another hidden factor. HTTP/HTTPS agents in Node.js default to 5 sockets per host. Under high concurrency, a burst of requests queues connections that eventually hit the queue timeout — surfaced as ETIMEDOUT even though individual connections to the host succeed. Setting maxSockets: Infinity on the agent (appropriate for short-lived Lambda processes) or increasing it to match your concurrency avoids queue-induced timeouts. The got and axios libraries expose separate connect, response, and overall timeout options that help distinguish which phase is timing out.