no_proxy plain domain entries don't match subdomains (inconsistent with Go/curl/Python behavior)

[vparfonov]

Question

Problem

Vector's no_proxy handling does not match subdomains when a plain domain (without leading dot) is specified. This is due to the upstream no-proxy crate (v0.3.6) which Vector uses for NO_PROXY evaluation.
When NO_PROXY=apps.example.com is set:

• Go's httpproxy, curl, Python, Ruby: elasticsearch.apps.example.com bypasses the proxy
• Vector: elasticsearch.apps.example.com goes through the proxy

Users must use .apps.example.com (with leading dot) as a workaround.

Root cause

The no-proxy crate's NoProxyItem::Plain variant does exact string match only: Self::Plain(source) => source == value,

Vector's documentation is also incorrect

In lib/vector-core/src/config/proxy.rs lines 83-84:

| Domain names     | `example.com` matches requests to `example.com`                     |
| Wildcard domains | `.example.com` matches requests to `example.com` and its subdomains |

Fix

Proposed a fix for the no-proxy crate: jdrouet/no-proxy#12

Vector Config

No response

Vector Logs

No response

[vparfonov]

UPD: proposed PR was merged and released in no-proxy: 0.3.7 https://github.com/jdrouet/no-proxy/releases/tag/v0.3.7