Crate trust_dns_resolver

Expand description

The Resolver is responsible for performing recursive queries to lookup domain names.

This is a 100% in process DNS resolver. It does not use the Host OS’ resolver. If what is desired is to use the Host OS’ resolver, generally in the system’s libc, then the std::net::ToSocketAddrs variant over &str should be used.

Unlike the trust-dns-client, this tries to provide a simpler interface to perform DNS queries. For update options, i.e. Dynamic DNS, the trust-dns-client crate must be used instead. The Resolver library is capable of searching multiple domains (this can be disabled by using an FQDN during lookup), dual-stack IPv4/IPv6 lookups, performing chained CNAME lookups, and features connection metric tracking for attempting to pick the best upstream DNS resolver.

There are two types for performing DNS queries, Resolver and AsyncResolver. Resolver is the easiest to work with, it is a wrapper around AsyncResolver. AsyncResolver is a Tokio based async resolver, and can be used inside any Tokio based system.

This as best as possible attempts to abide by the DNS RFCs, please file issues at https://github.com/bluejekyll/trust-dns.

Usage

Declare dependency

[dependency]
trust-dns-resolver = "*"

Using the Synchronous Resolver

This uses the default configuration, which sets the Google Public DNS as the upstream resolvers. Please see their privacy statement for important information about what they track, many ISP’s track similar information in DNS.

use std::net::*;
use trust_dns_resolver::Resolver;
use trust_dns_resolver::config::*;

// Construct a new Resolver with default configuration options
let resolver = Resolver::new(ResolverConfig::default(), ResolverOpts::default()).unwrap();

// Lookup the IP addresses associated with a name.
// The final dot forces this to be an FQDN, otherwise the search rules as specified
//  in `ResolverOpts` will take effect. FQDN's are generally cheaper queries.
let response = resolver.lookup_ip("www.example.com.").unwrap();

// There can be many addresses associated with the name,
//  this can return IPv4 and/or IPv6 addresses
let address = response.iter().next().expect("no addresses returned!");
if address.is_ipv4() {
    assert_eq!(address, IpAddr::V4(Ipv4Addr::new(93, 184, 216, 34)));
} else {
    assert_eq!(address, IpAddr::V6(Ipv6Addr::new(0x2606, 0x2800, 0x220, 0x1, 0x248, 0x1893, 0x25c8, 0x1946)));
}

Using the host system config

On Unix systems, the /etc/resolv.conf can be used for configuration. Not all options specified in the host systems resolv.conf are applicable or compatible with this software. In addition there may be additional options supported which the host system does not. Example:

// Use the host OS'es `/etc/resolv.conf`
let resolver = Resolver::from_system_conf().unwrap();
let response = resolver.lookup_ip("www.example.com.").unwrap();

Using the Tokio/Async Resolver

For more advanced asynchronous usage, the AsyncResolver] is integrated with Tokio. In fact, the AsyncResolver is used by the synchronous Resolver for all lookups.

use std::net::*;
use tokio::runtime::Runtime;
use trust_dns_resolver::TokioAsyncResolver;
use trust_dns_resolver::config::*;

// We need a Tokio Runtime to run the resolver
//  this is responsible for running all Future tasks and registering interest in IO channels
let mut io_loop = Runtime::new().unwrap();

// Construct a new Resolver with default configuration options
let resolver = io_loop.block_on(async {
    TokioAsyncResolver::tokio(
        ResolverConfig::default(),
        ResolverOpts::default())
}).expect("failed to connect resolver");

// Lookup the IP addresses associated with a name.
// This returns a future that will lookup the IP addresses, it must be run in the Core to
//  to get the actual result.
let lookup_future = resolver.lookup_ip("www.example.com.");

// Run the lookup until it resolves or errors
let mut response = io_loop.block_on(lookup_future).unwrap();

// There can be many addresses associated with the name,
//  this can return IPv4 and/or IPv6 addresses
let address = response.iter().next().expect("no addresses returned!");
if address.is_ipv4() {
    assert_eq!(address, IpAddr::V4(Ipv4Addr::new(93, 184, 216, 34)));
} else {
    assert_eq!(address, IpAddr::V6(Ipv6Addr::new(0x2606, 0x2800, 0x220, 0x1, 0x248, 0x1893, 0x25c8, 0x1946)));
}

Generally after a lookup in an asynchronous context, there would probably be a connection made to a server, for example:

let ips = io_loop.block_on(resolver.lookup_ip("www.example.com.")).unwrap();

let result = io_loop.block_on(async {
    let ip = ips.iter().next().unwrap();
    TcpStream::connect((ip, 443))
})
.and_then(|conn| Ok(conn) /* do something with the connection... */)
.unwrap();

It’s beyond the scope of these examples to show how to deal with connection failures and looping etc. But if you wanted to say try a different address from the result set after a connection failure, it will be necessary to create a type that implements the Future trait. Inside the Future::poll method would be the place to implement a loop over the different IP addresses.

DNS-over-TLS and DNS-over-HTTPS

DNS-over-TLS and DNS-over-HTTPS are supported in the Trust-DNS Resolver library. The underlying implementations are available as addon libraries. WARNING The trust-dns developers make no claims on the security and/or privacy guarantees of this implementation.

To use DNS-over-TLS one of the dns-over-tls features must be enabled at compile time. There are three: dns-over-openssl, dns-over-native-tls, and dns-over-rustls. For DNS-over-HTTPS only rustls is supported with the dns-over-https-rustls, this implicitly enables support for DNS-over-TLS as well. The reason for each is to make the Trust-DNS libraries flexible for different deployments, and/or security concerns. The easiest to use will generally be dns-over-rustls which utilizes the *ring* Rust cryptography library (a rework of the boringssl project), this should compile and be usable on most ARM and x86 platforms. dns-over-native-tls will utilize the hosts TLS implementation where available or fallback to openssl where not supported. dns-over-openssl will specify that openssl should be used (which is a perfectly fine option if required). If more than one is specified, the precedence will be in this order (i.e. only one can be used at a time) dns-over-rustls, dns-over-native-tls, and then dns-over-openssl. NOTICE the trust-dns developers are not responsible for any choice of library that does not meet required security requirements.

Example

Enable the TLS library through the dependency on trust-dns-resolver:

trust-dns-resolver = { version = "*", features = ["dns-over-rustls"] }

A default TLS configuration is available for Cloudflare’s 1.1.1.1 DNS service (Quad9 as well):

use trust_dns_resolver::Resolver;
use trust_dns_resolver::config::*;

// Construct a new Resolver with default configuration options
let mut resolver = Resolver::new(ResolverConfig::cloudflare_tls(), ResolverOpts::default()).unwrap();

// see example above...

mDNS (multicast DNS)

Multicast DNS is an experimental feature in Trust-DNS at the moment. Its support on different platforms is not yet ideal. Initial support is only for IPv4 mDNS, as there are some complexities to figure out with IPv6. Once enabled, an mDNS NameServer will automatically be added to the Resolver and used for any lookups performed in the .local. zone.