Node-named is a lightweight DNS server written in pure javascript. It has limited support for the DNS spec, but aims to implement all of the common functionality that is in use today.
var named = require('./lib/index');
var server = named.createServer();
var ttl = 300;
server.listen(9999, '127.0.0.1', function() {
console.log('DNS server started on port 9999');
});
server.on('query', function(query) {
var domain = query.name();
console.log('DNS Query: %s', domain)
var target = new named.SOARecord(domain, {serial: 12345});
query.addAnswer(domain, target, ttl);
server.send(query);
});
node-named provides helper functions for creating DNS records. The records are available under 'named.record.NAME' where NAME is one of ['A', 'AAAA', 'CNAME', 'SOA', 'MX', 'TXT, 'SRV']. It is important to remember that these DNS records are not permanently added to the server. They only exist fo the length of the particular request. After that, they are destroyed. This means you have to create your own lookup mechanism.
var named = require('node-named');
var soaRecord = named.SOARecord('example.com', {serial: 201205150000});
console.log(soaRecord);
The following record types are supported
- A (ipv4)
- AAAA (ipv6)
- CNAME (aliases)
- SOA (start of authority)
- MX (mail server records)
- TXT (arbitrary text entries)
- SRV (service discovery)
node-named uses http://github.com/trentm/node-bunyan for logging. It's a lot nicer to use if you npm install bunyan and put the bunyan tool in your path. Otherwise, you will end up with JSON formatted log output by default.
You can pass in an alternate logger if you wish. If you do not, then it will use bunyan by default. Your logger must expose the functions 'info', 'debug', 'warn', 'trace', 'error', and 'notice'.
- Better record validation
- Create DNS client for query recursor
- Add support for PTR records
- Add support for TCP AXFR requests
When DNS was designed it was designed prior to the web existing, so many of the features in the RFC are either never used, or were never implemented. This server aims to be RFC compliant, but does not implement any other protocol other than INET (the one we're all used to), and only supports a handful of record types (the ones that are in use on a regular basis).