Presence of a faulty nameserver in system config causes client side lookups to take forever

[bergentruckung]

Describe the bug
If there's a faulty nameserver (non-existent/typo or dummy) in /etc/resolv.conf along with others that are working, the lookup gets stalled (in my case, up to 16s). Would it be possible for the client side lookup to rely on the nameservers in order (from top to bottom), or have an alternate strategy of returning the first successful lookup?

To Reproduce

1. Add any non-existent IP to list of nameservers in /etc/resolv.conf
2. Try running resolve -s google.com

Expected behavior
DNS lookup to be quick.

System:

• OS: RHEL8
• Architecture: x86_64
• rustc version: [e.g. 1.73]

Version:
Crate: [e.g. client, server, resolver]
Version: [e.g. 0.24.0]

[djc]

It would be good to dig into more detail on this. This ends up calling AsyncResolver::inner_lookup(), but it would be good to understand what the ResolverConfig and ResolverOpts look like in this case. Unfortunately I won't be able to dig into this much more, but happy to provide guidance if you're able to provide more details on the behavior you're seeing.