Lambda Function Unable to Access the Internet Despite No VPC Configuration

The log errors come from dummy tests we've inserted in the function as simple as:

try { console.log(🔍 Testing https://httpbin.org/get...); const response = await axios.get("https://httpbin.org/get", { timeout: 5000 }); console.log(✅ SUCCESS for https://httpbin.org/get:, response.status); } catch (err) { console.error(❌ FAILED for https://httpbin.org/get:, err.message); }

in which we'll get

ERROR ❌ FAILED for https://httpbin.org/get: timeout of 5000ms exceeded

The Lambda timeout has been set to 29s which shouldn't be the issue since the execution normally completes <2s

As far as the permissions go, you're absolutely right. It was just sheer desperation at this point trying to figure out why a simple get will be blocked in our lambda function that doesn't have a VPC attached and should be able to access the internet

I think is probably related to the code itself and not AWS configurations. There is no reason of getting no internet access from your lambda function, outside of your VPC.

I've tested a very simple code, from yours :

import axios from "axios";

export const handler = async () => {
    try {
        console.log("🔍 Testing https://httpbin.org/get...");
        const response = await axios.get("https://httpbin.org/get", { timeout: 5000 });
        console.log("✅ SUCCESS for https://httpbin.org/get:", response.status);
        return {
            statusCode: 200,
            body: JSON.stringify(response.data),
        };
    } catch (err) {
        console.error("❌ FAILED for https://httpbin.org/get:", err.message);
        return {
            statusCode: 500,
            body: JSON.stringify({ error: err.message }),
        };
    }
};

It works well on my side. Have you tried using Axios in your main Lambda function with others configurations, or tried creating a simple and minimal Lambda function to test? Do you have other lambda which is working?

The code itself should be fine since it runs flawlessly on local (using ngrok tunneling). E2E tests with all external services passes. There's only an issue when we deploy this via serverless to lambda and specifically those issues (i.e. the app cannot seem to reach any external service). That's a good callout tho. I'll deploy a simple function to test and see if all lambdas are messed up in which case its probably some obscure account thing. I'll update!

Oh geez I'm such an idiot but I figured it out. See the invoker of the function has a pretty narrow timeout window so we respond immediately and dispatch the actual task async (and notify via callbacks and all that). It would appear for Lambdas that once you commit your response, AWS will forcibly kill your thread even if its running. There's no set time window to which it does that which was what made troubleshooting such a challenge since the logs were highly inconsistent. Sometimes the application would reach a certain stage and other times it gets terminated way before.

Pretty much means lambdas are not suitable for this use case I reckon. Thanks for your responses Brandon!

Ahah thank you for the feedback!
If I’ve helped you don’t forget to mark it as resolved, good luck in your project !