Error handling for EnvHttpProxyAgent

[reynaldichernando]

Hi team,
I want to ask about properly handling error in the case of the proxy URL (via HTTP_PROXY or HTTPS_PROXY) is unavailable.
Right now, the error message only gives basic "fetch failed" with cause: ECONNREFUSED (I guess it also includes address and port).

TypeError: fetch failed
    at fetch (/path-to-project/node_modules/.pnpm/[email protected]/node_modules/undici/index.js:123:13)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at (more stack trace)
{
  [cause]: Error: connect ECONNREFUSED 127.0.0.1:1080
      at TCPConnectWrap.afterConnect [as oncomplete] (node:net:1611:16) {
    errno: -61,
    code: 'ECONNREFUSED',
    syscall: 'connect',
    address: '127.0.0.1',
    port: 1080
  }
}

The reason I ask this because I want to discern between ECONNREFUSED when it is the target URL that is down or the proxy URL that is down.

---

Some info on how to reproduce:

.env

HTTP_PROXY=http://127.0.0.1:1080
HTTPS_PROXY=http://127.0.0.1:1080

(just have this set in the .env file, without any proxy running)

index.js

fetch("https://www.google.com", {
  dispatcher: new EnvHttpProxyAgent(),
});

Thank you.

[reynaldichernando]

I thought about using the address + port and checking the .env value if it is the proxy URL, but I'm not sure if this is the proper way.

Also, right now, the proxy URL is using an IP, having it as a domain/hostname will resolve the proxy URL to an actual IP, and this logic will fail, since the error cause address won't match the one in the .env.

[metcoder95]

You can implement a clientFactory when creating the new EnvHttpProxyAgent({ clientFactory }) and use a custom factory to create a proper client accordingly to your needs (e.g. extend it with traceability, or handle error and retry, etc.)