Javascript DSL

DNSControl uses javascript as its primary input language to provide power and flexibility to configure your domains. The ultimate purpose of the javascript is to construct a DNSConfig object that will be passed to the go backend and operated on.

Top Level Functions

require_glob( path, recursive )

require_glob() can recursively load .js files, optionally non-recursive as well.

Possible parameters are:

  • Path as string, where you would like to start including files. Mandatory. Pattern matching possible, see GoLand path/filepath/#Match docs.
  • If being recursive. This is a boolean if the search should be recursive or not. Define either true or false. Default is true.

Example to load .js files recursively:

require_glob("./domains/");

Example to load .js files only in domains/:

require_glob("./domains/", false);

One more important thing to note: require_glob() is as smart as require() is. It loads files always relative to the JavaScript file where it’s being executed in. Let’s go with an example, as it describes it better:

dnscontrol.js:

require("domains/index.js");

domains/index.js:

require_glob("./user1/");

This will now load files being present underneath ./domains/user1/ and NOT at below ./domains/, as require_glob() is called in the subfolder domains/.

require( path )

require(...) loads the specified JavaScript or JSON file, allowing to split your configuration across multiple files.

If the supplied path string ends with .js, the file is interpreted as JavaScript code, almost as though its contents had been included in the currently-executing file. If the path string ends with .json, require() returns the JSON.parse() of the file’s contents.

If the path string begins with a ., it is interpreted relative to the currently-loading file (which may not be the file where the require() statement is, if called within a function), otherwise it is interpreted relative to the program’s working directory at the time of the call.

Example:

// dnsconfig.js
require('kubernetes/clusters.js');

D("mydomain.net", REG, PROVIDER,
    IncludeKubernetes()
);
// kubernetes/clusters.js
require('./clusters/prod.js');
require('./clusters/dev.js');

function IncludeKubernetes() {
    return [includeK8Sprod(), includeK8Sdev()];
}
// kubernetes/clusters/prod.js
function includeK8Sprod() {
    return [ /* ... */ ];
}
// kubernetes/clusters/dev.js
function includeK8Sdev() {
    return [ /* ... */ ];
}

You can also use it to require json files and initialize variables with it: For Example:

Example:

// dnsconfig.js
var domains = require('./domain-ip-map.json')

for (var domain in domains) {
    D(domain, REG, PROVIDER,
        A("@", domains[domain])
    );
}
// domain-ip-map.json
{
    "mydomain.net": "1.1.1.1",
    "myotherdomain.org": "5.5.5.5"
}
# Future It might be better to rename the function to something like `include()` instead, (leaving `require` as a deprecated alias) because by analogy it is *much* closer to PHP's `include()` function than it is to node's `require()`. After all, the reason node.js calls it "require" is because it's a declarative statement saying the file is needed, and so should be loaded if it hasn't already been loaded. In contrast, dnscontrol's require is actually an imperative command to load the file and execute the code or parse the data from it. (So if two files both `require("./tools.js")`, for example, then it will be loaded twice, whereas in node.js it would only be loaded once.)

getConfiguredDomains( name, modifiers... )

getConfiguredDomains getConfiguredDomains is a helper function that returns the domain names configured at the time the function is called. Calling this function early or later in dnsconfig.js may return different results. Typical usage is to iterate over all domains at the end of your configuration file.

Example for adding records to all configured domains: Example:

var domains = getConfiguredDomains();
for(i = 0; i < domains.length; i++) {
  D_EXTEND(domains[i],
    TXT('_important', 'BLA') // I know, not really creative.
  )
}
This will end up in following modifications: ``` ******************** Domain: domain1.tld ----- Getting nameservers from: registrar ----- DNS Provider: registrar...2 corrections #1: CREATE TXT _important.domain1.tld "BLA" ttl=43200 #2: REFRESH zone domain1.tld ******************** Domain: domain2.tld ----- Getting nameservers from: registrar ----- DNS Provider: registrar...2 corrections #1: CREATE TXT _important.domain2.tld "BLA" ttl=43200 #2: REFRESH zone domain2.tld ```

Example for adding DMARC report records: Example:

This example might be more useful, specially for configuring the DMARC report records. According to DMARC RFC you need to specify `domain2.tld._report.dmarc.domain1.tld` to allow `domain2.tld` to send aggregate/forensic email reports to `domain1.tld`. This can be used to do this in an easy way, without using the wildcard from the RFC.
var domains = getConfiguredDomains();
for(i = 0; i < domains.length; i++) {
	D_EXTEND("domain1.tld",
		TXT(domains[i] + '._report._dmarc', 'v=DMARC1')
	);
}
This will end up in following modifications: ``` ******************** Domain: domain2.tld ----- Getting nameservers from: registrar ----- DNS Provider: registrar...4 corrections #1: CREATE TXT domain1.tld._report._dmarc.domain2.tld "v=DMARC1" ttl=43200 #2: CREATE TXT domain3.tld._report._dmarc.domain2.tld "v=DMARC1" ttl=43200 #3: CREATE TXT domain4.tld._report._dmarc.domain2.tld "v=DMARC1" ttl=43200 #4: REFRESH zone domain2.tld ```

REV( address )

REV returns the reverse lookup domain for an IP network. For example REV('1.2.3.0/24') returns 3.2.1.in-addr.arpa. and REV('2001:db8:302::/48) returns 2.0.3.0.8.b.d.0.1.0.0.2.ip6.arpa.. This is used in D() functions to create reverse DNS lookup zones.

This is a convenience function. You could specify D('3.2.1.in-addr.arpa, …` if you like to do things manually but why would you risk making typos?

REV complies with RFC2317, “Classless in-addr.arpa delegation” for netmasks of size /25 through /31. While the RFC permits any format, we abide by the recommended format: FIRST/MASK.C.B.A.in-addr.arpa where FIRST is the first IP address of the zone, MASK is the netmask of the zone (25-31 inclusive), and A, B, C are the first 3 octets of the IP address. For example 172.20.18.130/27 is located in a zone named 128/27.18.20.172.in-addr.arpa

If the address does not include a “/” then REV assumes /32 for IPv4 addresses and /128 for IPv6 addresses.

Note that the lower bits (the ones outside the netmask) must be zeros. They are not zeroed out automatically. Thus, `REV(‘1.2.3.4/24’) is an error. This is done to catch typos.

Example:

D(REV('1.2.3.0/24'), REGISTRAR, DnsProvider(BIND),
  PTR("1", 'foo.example.com.'),
  PTR("2", 'bar.example.com.'),
  PTR("3", 'baz.example.com.'),
  // These take advantage of DNSControl's ability to generate the right name:
  PTR("1.2.3.10", 'ten.example.com.'),
);

D(REV('2001:db8:302::/48'), REGISTRAR, DnsProvider(BIND),
  PTR("1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0", 'foo.example.com.'),  // 2001:db8:302::1
  // These take advantage of DNSControl's ability to generate the right name:
  PTR("2001:db8:302::2", 'two.example.com.'),                          // 2.0.0...
  PTR("2001:db8:302::3", 'three.example.com.'),                        // 3.0.0...
);

In the future we plan on adding a flag to A() which will insert the correct PTR() record if the appropriate D(REV() domain (i.e. .arpa domain) has been defined.

PANIC( message )

PANIC terminates the script and therefore DnsControl with an exit code of 1. This should be used if your script cannot gather enough information to generate records, for example when a HTTP request failed.

Example:

PANIC("Something really bad has happened");

NewRegistrar( name, type, meta ) -> string

NewRegistrar registers a registrar provider. The name can be any string value you would like to use. The type must match a valid registrar provider type identifier (see provider page.)

Metadata is an optional object, that will only be used by certain providers. See individual provider docs for specific details.

This function will return the name as a string so that you may assign it to a variable to use inside D directives.

Example:

var REGISTRAR = NewRegistrar("name.com", "NAMEDOTCOM");
var r53 = NewDnsProvider("R53","ROUTE53");

D("example.com", REGISTRAR, DnsProvider(r53), A("@","1.2.3.4"));

NewDnsProvider( name, type, meta ) -> string

NewDnsProvider registers a new DNS Service Provider. The name can be any string value you would like to use. The type must match a valid dns provider type identifier (see provider page.)

Metadata is an optional object, that will only be used by certain providers. See individual provider docs for specific details.

This function will return the name as a string so that you may assign it to a variable to use inside D directives.

Example:

var REGISTRAR = NewRegistrar("name.com", "NAMEDOTCOM");
var R53 = NewDnsProvider("r53", "ROUTE53");

D("example.com", REGISTRAR, DnsProvider(R53), A("@","1.2.3.4"));

IP( ip )

Converts an IPv4 address from string to an integer. This allows performing mathematical operations with the IP address.

This does not accept IPv6 addresses. (PRs gladly accepted.)

Example:

var addrA = IP('1.2.3.4')
var addrB = addrA + 1
// addrB = 1.2.3.5

FETCH( url, args )

FETCH is a wrapper for the Fetch API. This allows dynamically setting DNS records based on an external data source, e.g. the API of your cloud provider.

Compared to fetch from Fetch API, FETCH will call PANIC to terminate the execution of the script, and therefore DnsControl, if a network error occurs.

Otherwise the syntax of FETCH is the same as fetch.

FETCH is not enabled by default. Please read the warnings below.

WARNING:

  1. Relying on external sources adds a point of failure. If the external source doesn’t work, your script won’t either. Please make sure you are aware of the consequences.
  2. Make sure DnsControl only uses verified configuration if you want to use FETCH. For example, an attacker can send Pull Requests to your config repo, and have your CI test malicious configurations and make arbitrary HTTP requests. Therefore, FETCH must be explicitly enabled with flag --allow-fetch on DnsControl invocation.

Example:

var REG_NONE = NewRegistrar('none', 'NONE');
var DNS_BIND = NewDnsProvider('bind', 'BIND');

D('example.com', REG_NONE, DnsProvider(DNS_BIND), [
  A('@', '1.2.3.4'),
]);

FETCH('https://example.com', {
  // All three options below are optional
  headers: {"X-Authentication": "barfoo"},
  method: "POST",
  body: "Hello World",
}).then(function(r) {
  return r.text();
}).then(function(t) {
  // Example of generating record based on response
  D_EXTEND('example.com', [
    TXT('@', t.slice(0, 100)),
  ]);
});

D_EXTEND( name, modifiers... )

D_EXTEND adds records (and metadata) to a domain previously defined by D(). It can also be used to add subdomain records (and metadata) to a previously defined domain.

The first argument is a domain name. If it exactly matches a previously defined domain, D_EXTEND() behaves the same as D(), simply adding records as if they had been specified in the original D().

If the domain name does not match an existing domain, but could be a (non-delegated) subdomain of an existing domain, the new records (and metadata) are added with the subdomain part appended to all record names (labels), and targets (as appropriate). See the examples below.

Matching the domain name to previously-defined domains is done using a longest match algorithm. If domain.tld and sub.domain.tld are defined as separate domains via separate D() statements, then D_EXTEND('sub.sub.domain.tld', ...) would match sub.domain.tld, not domain.tld.

Some operators only act on an apex domain (e.g. CF_REDIRECT and CF_TEMP_REDIRECT). Using them in a D_EXTEND subdomain may not be what you expect.

Example:

Example:

D("domain.tld", REG, DnsProvider(DNS),
  A("@", "127.0.0.1"), // domain.tld
  A("www", "127.0.0.2"), // www.domain.tld
  CNAME("a", "b") // a.domain.tld -> b.domain.tld
);
D_EXTEND("domain.tld",
  A("aaa", "127.0.0.3"), // aaa.domain.tld
  CNAME("c", "d") // c.domain.tld -> d.domain.tld
);
D_EXTEND("sub.domain.tld",
  A("bbb", "127.0.0.4"), // bbb.sub.domain.tld
  A("ccc", "127.0.0.5"), // ccc.sub.domain.tld
  CNAME("e", "f") // e.sub.domain.tld -> f.sub.domain.tld
);
D_EXTEND("sub.sub.domain.tld",
  A("ddd", "127.0.0.6"), // ddd.sub.sub.domain.tld
  CNAME("g", "h") // g.sub.sub.domain.tld -> h.sub.sub.domain.tld
);
D_EXTEND("sub.domain.tld",
  A("@", "127.0.0.7"), // sub.domain.tld
  CNAME("i", "j") // i.sub.domain.tld -> j.sub.domain.tld
);
This will end up in the following modifications: ``` ******************** Domain: domain.tld ----- Getting nameservers from: cloudflare ----- DNS Provider: cloudflare...7 corrections #1: CREATE A aaa.domain.tld 127.0.0.3 #2: CREATE A bbb.sub.domain.tld 127.0.0.4 #3: CREATE A ccc.sub.domain.tld 127.0.0.5 #4: CREATE A ddd.sub.sub.domain.tld 127.0.0.6 #5: CREATE A sub.domain.tld 127.0.0.7 #6: CREATE A www.domain.tld 127.0.0.2 #7: CREATE A domain.tld 127.0.0.1 #8: CREATE CNAME a.domain.tld b.domain.tld. #9: CREATE CNAME c.domain.tld d.domain.tld. #10: CREATE CNAME e.sub.domain.tld f.sub.domain.tld. #11: CREATE CNAME g.sub.sub.domain.tld h.sub.sub.domain.tld. #12: CREATE CNAME i.sub.domain.tld j.sub.domain.tld. ```

ProTips: D_EXTEND() permits you to create very complex and sophisticated configurations, but you shouldn’t. Be nice to the next person that edits the file, who may not be as expert as yourself. Enhance readability by putting any D_EXTEND() statements immediately after the original D(), like in above example. Avoid the temptation to obscure the addition of records to existing domains with randomly placed D_EXTEND() statements. Don’t build up a domain using loops of D_EXTEND() statements. You’ll be glad you didn’t.

DEFAULTS( modifiers... )

DEFAULTS allows you to declare a set of default arguments to apply to all subsequent domains. Subsequent calls to D will have these arguments passed as if they were the first modifiers in the argument list.

Example:

var COMMON = NewDnsProvider("foo","BIND");
// we want to create backup zone files for all domains, but not actually register them.
// also create a default TTL
DEFAULTS( DnsProvider(COMMON,0), DefaultTTL(1000));

D("example.com", REGISTRAR, DnsProvider("R53"), A("@","1.2.3.4")); // this domain will have the defaults set.

// clear defaults
DEFAULTS();
D("example2.com", REGISTRAR, DnsProvider("R53"), A("@","1.2.3.4")); // this domain will not have the previous defaults.

D( name, registrar, modifiers... )

D adds a new Domain for DNSControl to manage. The first two arguments are required: the domain name (fully qualified example.com without a trailing dot), and the name of the registrar (as previously declared with NewRegistrar). Any number of additional arguments may be included to add DNS Providers with DNSProvider, add records with A, CNAME, and so forth, or add metadata.

Modifier arguments are processed according to type as follows:

  • A function argument will be called with the domain object as it’s only argument. Most of the built-in modifier functions return such functions.
  • An object argument will be merged into the domain’s metadata collection.
  • An array argument will have all of it’s members evaluated recursively. This allows you to combine multiple common records or modifiers into a variable that can be used like a macro in multiple domains.

Example:

var REGISTRAR = NewRegistrar("name.com", "NAMEDOTCOM");
var r53 = NewDnsProvider("R53","ROUTE53");

// simple domain
D("example.com", REGISTRAR, DnsProvider(r53),
  A("@","1.2.3.4"),
  CNAME("test", "foo.example2.com.")
);

// "macro" for records that can be mixed into any zone
var GOOGLE_APPS_DOMAIN_MX = [
    MX('@', 1, 'aspmx.l.google.com.'),
    MX('@', 5, 'alt1.aspmx.l.google.com.'),
    MX('@', 5, 'alt2.aspmx.l.google.com.'),
    MX('@', 10, 'alt3.aspmx.l.google.com.'),
    MX('@', 10, 'alt4.aspmx.l.google.com.'),
]

D("example.com", REGISTRAR, DnsProvider(r53),
  A("@","1.2.3.4"),
  CNAME("test", "foo.example2.com."),
  GOOGLE_APPS_DOMAIN_MX
);

Split Horizon DNS

DNSControl supports Split Horizon DNS. Simply define the domain two or more times, each with their own unique parameters.

To differentiate the different domains, specify the domains as domain.tld!tag, such as example.com!inside and example.com!outside.

Example:

var REG = NewRegistrar("Third-Party", "NONE");
var DNS_INSIDE = NewDnsProvider("Cloudflare", "CLOUDFLAREAPI");
var DNS_OUTSIDE = NewDnsProvider("bind", "BIND");

D("example.com!inside", REG, DnsProvider(DNS_INSIDE),
  A("www", "10.10.10.10")
);

D("example.com!outside", REG, DnsProvider(DNS_OUTSIDE),
  A("www", "20.20.20.20")
);

D_EXTEND("example.com!inside",
  A("internal", "10.99.99.99")
);

A domain name without a ! is assigned a tag that is the empty string. For example, example.com and example.com! are equivalent. However, we strongly recommend against using the empty tag, as it risks creating confusion. In other words, if you have domain.tld and domain.tld!external you now require humans to remember that domain.tld is the external one. I mean… the internal one. You may have noticed this mistake, but will your coworkers? Will you in six months? You get the idea.

DNSControl command line flag --domains is an exact match. If you define domains example.com!george and example.com!john then:

  • --domains=example.com will not match either domain.
  • --domains='example.com!george' will match only match the first.
  • --domains='example.com!george',example.com!john will match both.

NOTE: The quotes are required if your shell treats ! as a special character, which is probably does. If you see an error that mentions event not found you probably forgot the quotes.

Domain Modifiers

A( name, address, modifiers... )

A adds an A record To a domain. The name should be the relative label for the record. Use @ for the domain apex.

The address should be an ip address, either a string, or a numeric value obtained via IP.

Modifiers can be any number of record modifiers or json objects, which will be merged into the record’s metadata.

Example:

D("example.com", REGISTRAR, DnsProvider("R53"),
  A("@", "1.2.3.4"),
  A("foo", "2.3.4.5"),
  A("test.foo", IP("1.2.3.4"), TTL(5000)),
  A("*", "1.2.3.4", {foo: 42})
);

AAAA( name, address, modifiers... )

AAAA adds an AAAA record To a domain. The name should be the relative label for the record. Use @ for the domain apex.

The address should be an IPv6 address as a string.

Modifiers can be any number of record modifiers or json objects, which will be merged into the record’s metadata.

Example:

var addrV6 = "2001:0db8:85a3:0000:0000:8a2e:0370:7334"

D("example.com", REGISTRAR, DnsProvider("R53"),
  AAAA("@", addrV6),
  AAAA("foo", addrV6),
  AAAA("test.foo", addrV6, TTL(5000)),
  AAAA("*", addrV6, {foo: 42})
);

AKAMAICDN( name, target, modifiers... )

AKAMAICDN is a proprietary record type that is used to configure Zone Apex Mapping. The AKAMAICDN target must be preconfigured in the Akamai network.

ALIAS( name, target, modifiers... )

ALIAS is a virtual record type that points a record at another record. It is analogous to a CNAME, but is usually resolved at request-time and served as an A record. Unlike CNAMEs, ALIAS records can be used at the zone apex (@)

Different providers handle ALIAS records differently, and many do not support it at all. Attempting to use ALIAS records with a DNS provider type that does not support them will result in an error.

The name should be the relative label for the domain.

Target should be a string representing the target. If it is a single label we will assume it is a relative name on the current domain. If it contains any dots, it should be a fully qualified domain name, ending with a ..

Example:

D("example.com", REGISTRAR, DnsProvider("CLOUDFLARE"),
  ALIAS("@", "google.com."), // example.com -> google.com
);

AUTODNSSEC_OFF( )

AUTODNSSEC_OFF tells the provider to disable AutoDNSSEC. It takes no parameters.

See AUTODNSSEC_ON for further details.

AUTODNSSEC_ON( )

AUTODNSSEC_ON tells the provider to enable AutoDNSSEC.

AUTODNSSEC_OFF tells the provider to disable AutoDNSSEC.

AutoDNSSEC is a feature where a DNS provider can automatically manage DNSSEC for a domain. Not all providers support this.

At this time, AUTODNSSEC_ON takes no parameters. There is no ability to tune what the DNS provider sets, no algorithm choice. We simply ask that they follow their defaults when enabling a no-fuss DNSSEC data model.

NOTE: No parenthesis should follow these keywords. That is, the correct syntax is AUTODNSSEC_ON not AUTODNSSEC_ON()

Example:

D("example.com", .... ,
  AUTODNSSEC_ON,  // Enable AutoDNSSEC.
  A("@", "10.1.1.1")
);

D("insecure.com", .... ,
  AUTODNSSEC_OFF,  // Disable AutoDNSSEC.
  A("@", "10.2.2.2")
);

If neither AUTODNSSEC_ON or AUTODNSSEC_OFF is specified for a domain no changes will be requested.

AZURE_ALIAS( name, type, target, modifiers ... )

AZURE_ALIAS is a Azure specific virtual record type that points a record at either another record or an Azure entity. It is analogous to a CNAME, but is usually resolved at request-time and served as an A record. Unlike CNAMEs, ALIAS records can be used at the zone apex (@)

Unlike the regular ALIAS directive, AZURE_ALIAS is only supported on AZURE. Attempting to use AZURE_ALIAS on another provider than Azure will result in an error.

The name should be the relative label for the domain.

The type can be any of the following:

  • A
  • AAAA
  • CNAME

Target should be the Azure Id representing the target. It starts /subscription/. The resource id can be found in https://resources.azure.com/.

The Target can :

  • Point to a public IP resource from a DNS A/AAAA record set. You can create an A/AAAA record set and make it an alias record set to point to a public IP resource (standard or basic). The DNS record set changes automatically if the public IP address changes or is deleted. Dangling DNS records that point to incorrect IP addresses are avoided. There is a current limit of 20 alias records sets per resource.
  • Point to a Traffic Manager profile from a DNS A/AAAA/CNAME record set. You can create an A/AAAA or CNAME record set and use alias records to point it to a Traffic Manager profile. It’s especially useful when you need to route traffic at a zone apex, as traditional CNAME records aren’t supported for a zone apex. For example, say your Traffic Manager profile is myprofile.trafficmanager.net and your business DNS zone is contoso.com. You can create an alias record set of type A/AAAA for contoso.com (the zone apex) and point to myprofile.trafficmanager.net.
  • Point to an Azure Content Delivery Network (CDN) endpoint. This is useful when you create static websites using Azure storage and Azure CDN.
  • Point to another DNS record set within the same zone. Alias records can reference other record sets of the same type. For example, a DNS CNAME record set can be an alias to another CNAME record set. This arrangement is useful if you want some record sets to be aliases and some non-aliases.

Example:

D("example.com", REGISTRAR, DnsProvider("AZURE_DNS"),
  AZURE_ALIAS("foo", "A", "/subscriptions/726f8cd6-6459-4db4-8e6d-2cd2716904e2/resourceGroups/test/providers/Microsoft.Network/trafficManagerProfiles/testpp2"), // record for traffic manager
  AZURE_ALIAS("foo", "CNAME", "/subscriptions/726f8cd6-6459-4db4-8e6d-2cd2716904e2/resourceGroups/test/providers/Microsoft.Network/dnszones/example.com/A/quux."), // record in the same zone
);

CAA( name, tag, value, modifiers... )

CAA adds a CAA record to a domain. The name should be the relative label for the record. Use @ for the domain apex.

Tag can be one of “issue”, “issuewild” or “iodef”.

Value is a string. The format of the contents is different depending on the tag. DNSControl will handle any escaping or quoting required, similar to TXT records. For example use CAA("@", "issue", "letsencrypt.org") rather than CAA("@", "issue", "\"letsencrypt.org\"").

Flags are controlled by modifier:

  • CAA_CRITICAL: Issuer critical flag. CA that does not understand this tag will refuse to issue certificate for this domain.

CAA record is supported only by BIND, Google Cloud DNS, Amazon Route 53 and OVH. Some certificate authorities may not support this record until the mandatory date of September 2017.

Example:

D("example.com", REGISTRAR, DnsProvider("GCLOUD"),
  // Allow letsencrypt to issue certificate for this domain
  CAA("@", "issue", "letsencrypt.org"),
  // Allow no CA to issue wildcard certificate for this domain
  CAA("@", "issuewild", ";"),
  // Report all violation to test@example.com. If CA does not support
  // this record then refuse to issue any certificate
  CAA("@", "iodef", "mailto:test@example.com", CAA_CRITICAL)
);

CF_REDIRECT( destination, modifiers... )

CF_REDIRECT uses Cloudflare-specific features (“Forwarding URL” Page Rules) to generate a HTTP 301 permanent redirect.

If any CF_REDIRECT or CF_TEMP_REDIRECT functions are used then dnscontrol will manage all “Forwarding URL” type Page Rules for the domain. Page Rule types other than “Forwarding URL” will be left alone.

WARNING: Cloudflare does not currently fully document the Page Rules API and this interface is not extensively tested. Take precautions such as making backups and manually verifying dnscontrol preview output before running dnscontrol push. This is especially true when mixing Page Rules that are managed by DNSControl and those that aren’t.

HTTP 301 redirects are cached by browsers forever, usually ignoring any TTLs or other cache invalidation techniques. It should be used with great care. We suggest using a CF_TEMP_REDIRECT initially, then changing to a CF_REDIRECT only after sufficient time has elapsed to prove this is what you really want.

This example redirects the bare (aka apex, or naked) domain to www:

Example:

D("foo.com", .... ,
  CF_REDIRECT("mydomain.com/*", "https://www.mydomain.com/$1"),
);

CF_TEMP_REDIRECT( destination, modifiers... )

CF_TEMP_REDIRECT uses Cloudflare-specific features (“Forwarding URL” Page Rules) to generate a HTTP 302 temporary redirect.

If any CF_REDIRECT or CF_TEMP_REDIRECT functions are used then dnscontrol will manage all “Forwarding URL” type Page Rules for the domain. Page Rule types other than “Forwarding URL” will be left alone.

WARNING: Cloudflare does not currently fully document the Page Rules API and this interface is not extensively tested. Take precautions such as making backups and manually verifying dnscontrol preview output before running dnscontrol push. This is especially true when mixing Page Rules that are managed by DNSControl and those that aren’t.

Example:

D("foo.com", .... ,
  CF_TEMP_REDIRECT("example.mydomain.com/*", "https://otherplace.yourdomain.com/$1"),
);

CNAME( name, target, modifiers... )

CNAME adds a CNAME record to the domain. The name should be the relative label for the domain. Using @ or * for CNAME records is not recommended, as different providers support them differently.

Target should be a string representing the CNAME target. If it is a single label we will assume it is a relative name on the current domain. If it contains any dots, it should be a fully qualified domain name, ending with a ..

Example:

D("example.com", REGISTRAR, DnsProvider("R53"),
  CNAME("foo", "google.com."), // foo.example.com -> google.com
  CNAME("abc", "@"), // abc.example.com -> example.com
  CNAME("def", "test"), // def.example.com -> test.example.com
);

DS( name, keytag, algorithm, digesttype, digest, modifiers... )

DS adds a DS record to the domain.

Key Tag should be a number.

Algorithm should be a number.

Digest Type must be a number.

Digest must be a string.

Example:

D("example.com", REGISTRAR, DnsProvider(R53),
  DS("example.com", 2371, 13, 2, "ABCDEF")
);

DefaultTTL( ttl )

DefaultTTL sets the TTL for all records in a domain that do not explicitly set one with TTL. If neither DefaultTTl or TTL exist for a record, it will use the DNSControl global default of 300 seconds.

Example:

D('example.com', REGISTRAR, DnsProvider('R53'),
  DefaultTTL("4h"),
  A('@','1.2.3.4'), // uses default
  A('foo', '2.3.4.5', TTL(600)) // overrides default
);
The DefaultTTL duration is the same format as [TTL](#TTL), an integer number of seconds or a string with a unit such as `'4d'`.

DnsProvider( name, nsCount )

DnsProvider indicates that the specified provider should be used to manage records for this domain. The name must match the name used with NewDnsProvider.

The nsCount parameter determines how the nameservers will be managed from this provider.

Leaving the parameter out means “fetch and use all nameservers from this provider as authoritative”. ie: DnsProvider("name")

Using 0 for nsCount means “do not fetch nameservers from this domain, or give them to the registrar”.

Using a different number, ie: DnsProvider("name",2), means “fetch all nameservers from this provider, but limit it to this many.

See this page for a detailed explanation of how DNSControl handles nameservers and NS records.

If a domain (D()) does not include any DnsProvider() functions, the DNS records will not be modified. In fact, if you want to control the Registrar for a domain but not the DNS records themselves, simply do not include a DnsProvider() function for that D().

FRAME( name, target, modifiers... )

Documentation needed.

IGNORE( )

IGNORE has been renamed to IGNORE_NAME. IGNORE will continue to function, but its use is deprecated. Please update your configuration files to use IGNORE_NAME.

( )

WARNING: The IGNORE_* family of functions is risky to use. The code is brittle and has subtle bugs. Use at your own risk. Do not use these commands with D_EXTEND().

IGNORE_NAME can be used to ignore some records present in zone. All records (independently of their type) of that name will be completely ignored.

IGNORE_NAME is like NO_PURGE except it acts only on some specific records instead of the whole zone.

Technically IGNORE_NAME is a promise that DNSControl will not add, change, or delete records at a given label. This permits another entity to “own” that label.

IGNORE_NAME is generally used in very specific situations:

  • Some records are managed by some other system and DNSControl is only used to manage some records and/or keep them updated. For example a DNS record that is managed by Kubernetes External DNS, but DNSControl is used to manage the rest of the zone. In this case we don’t want DNSControl to try to delete the externally managed record.
  • To work-around a pseudo record type that is not supported by DNSControl. For example some providers have a fake DNS record type called “URL” which creates a redirect. DNSControl normally deletes these records because it doesn’t understand them. IGNORE_NAME will leave those records alone.

In this example, DNSControl will insert/update the “baz.example.com” record but will leave unchanged the “foo.example.com” and “bar.example.com” ones.

Example:

D("example.com",
  `IGNORE_NAME`("foo"),
  `IGNORE_NAME`("bar"),
  A("baz", "1.2.3.4")
);

IGNORE_NAME also supports glob patterns in the style of the gobwas/glob library. All of the following patterns will work:

  • IGNORE_NAME("*.foo") will ignore all records in the style of bar.foo, but will not ignore records using a double subdomain, such as foo.bar.foo.
  • IGNORE_NAME("**.foo") will ignore all subdomains of foo, including double subdomains.
  • IGNORE_NAME("?oo") will ignore all records of three symbols ending in oo, for example foo and zoo. It will not match .
  • IGNORE_NAME("[abc]oo") will ignore records aoo, boo and coo. IGNORE_NAME("[a-c]oo") is equivalent.
  • IGNORE_NAME("[!abc]oo") will ignore all three symbol records ending in oo, except for aoo, boo, coo. IGNORE_NAME("[!a-c]oo") is equivalent.
  • IGNORE_NAME("{bar,[fz]oo}") will ignore bar, foo and zoo.
  • IGNORE_NAME("\\*.foo") will ignore the literal record *.foo.

Caveats

It is considered as an error to try to manage an ignored record. Ignoring a label is a promise that DNSControl won’t meddle with anything at a particular label, therefore DNSControl prevents you from adding records at a label that is IGNORE_NAME‘ed.

Use IGNORE_NAME("@") to ignore at the domain’s apex. Most providers insert magic or unchangable records at the domain’s apex; usually NS and SOA records. DNSControl treats them specially.

Errors

  • trying to update/add IGNORE_NAME'd record: foo CNAME

This means you have both ignored foo and included a record (in this case, a CNAME) to update it. This is an error because IGNORE_NAME is a promise not to modify records at a certain label so that others may have free reign there. Therefore, DNSControl prevents you from modifying that label.

The foo CNAME at the end of the message indicates the label name (foo) and the type of record (CNAME) that your dnsconfig.js file is trying to insert.

You can override this error by adding the IGNORE_NAME_DISABLE_SAFETY_CHECK flag to the record.

TXT('vpn', "this thing", IGNORE_NAME_DISABLE_SAFETY_CHECK)

Disabling this safety check creates two risks:

  1. Two owners (DNSControl and some other entity) toggling a record between two settings.
  2. The other owner wiping all records at this label, which won’t be noticed until the next time dnscontrol is run.

IGNORE_TARGET( pattern, rType )

WARNING: The IGNORE_* family of functions is risky to use. The code is brittle and has subtle bugs. Use at your own risk. Do not use these commands with D_EXTEND() or use it at the domain apex.

IGNORE_TARGET can be used to ignore some records present in zone based on the record’s target and type. IGNORE_TARGET currently only supports CNAME record types.

IGNORE_TARGET is like NO_PURGE except it acts only on some specific records instead of the whole zone.

IGNORE_TARGET is generally used in very specific situations:

  • Some records are managed by some other system and DNSControl is only used to manage some records and/or keep them updated. For example a DNS record that is created by AWS Certificate Manager for validation, but DNSControl is used to manage the rest of the zone. In this case we don’t want DNSControl to try to delete the externally managed record.

In this example, DNSControl will insert/update the “baz.example.com” record but will leave unchanged a CNAME to “foo.acm-validations.aws” record.

Example:

D("example.com",
  IGNORE_TARGET('**.acm-validations.aws.', 'CNAME'),
  A("baz", "1.2.3.4")
);

IGNORE_TARGET also supports glob patterns in the style of the gobwas/glob library. Some example patterns:

  • IGNORE_TARGET("example.com", "CNAME") will ignore all CNAME records with targets of exactly example.com.
  • IGNORE_TARGET("*.foo", "CNAME") will ignore all CNAME records with targets in the style of bar.foo, but will not ignore records with targets using a double subdomain, such as foo.bar.foo.
  • IGNORE_TARGET("**.bar", "CNAME") will ignore all CNAME records with target subdomains of bar, including double subdomains such as www.foo.bar.
  • IGNORE_TARGET("dev.*.foo", "CNAME") will ignore all CNAME records with targets in the style of dev.bar.foo, but will not ignore records with targets using a double subdomain, such as dev.foo.bar.foo.

It is considered as an error to try to manage an ignored record.

IMPORT_TRANSFORM( transform table, domain, modifiers... )

Don’t use this feature. It was added for a very specific situation at Stack Overflow.

IMPORT_TRANSFORM adds to the domain all the records from another domain, after making certain transformations and resetting the TTL.

Example:

Suppose foo.com is a regular domain. bar.com is a regular domain, but certain records should be the same as foo.com with these exceptions: “bar.com” is added to the name, the TTL is changed to 300, if the IP address is between 1.2.3.10 and 1.2.3.20 then rewrite the IP address to be based on 123.123.123.100 (i.e. .113 or .114).

You wouldn’t want to maintain bar.com manually, would you? It would be very error prone. Therefore instead you maintain foo.com and let IMPORT_TRANSFORM automatically generate bar.com.

Example:

foo.com:
    one.foo.com.    IN A 1.2.3.1
    two.foo.com.    IN A 1.2.3.2
    three.foo.com.  IN A 1.2.3.13
    four.foo.com.   IN A 1.2.3.14

bar.com:
    www.bar.com.    IN 123.123.123.123
    one.foo.com.bar.com.    IN A 1.2.3.1
    two.foo.com.bar.com.    IN A 1.2.3.2
    three.foo.com.bar.com.  IN A 123.123.123.113
    four.foo.com.bar.com.   IN A 123.123.123.114

Here’s how you’d implement this in DNSControl:

Example:

var TRANSFORM_INT = [
    // RANGE_START, RANGE_END, NEW_BASE
    { low: "1.2.3.10", high: "1.2.3.20", newBase: "123.123.123.100" },  //   .10 to .20 rewritten as 123.123.123.100+IP
    { low: "2.4.6.80", high: "2.4.6.90", newBase: "123.123.123.200" },  //   Another rule, just to show that you can have many.
]

D("foo.com", .... ,
  A("one","1.2.3.1")
  A("two","1.2.3.2")
  A("three","1.2.3.13")
  A("four","1.2.3.14")
);

D("bar.com", .... ,
  A("www","123.123.123.123")
  IMPORT_TRANSFORM(TRANSFORM_INT, 'foo.com', 300),
);

Transform rules are: RANGE_START, RANGE_END, NEW_BASE. NEW_BASE may be:

  • An IP address. Rebase the IP address on this IP address. Extract the host part of the /24 and add it to the “new base” address.
  • A list of IP addresses. For each A record, inject an A record for each item in the list: newBase: ['1.2.3.100', '2.4.6.8.100'] would produce 2 records for each A record.

INCLUDE( domain )

Includes all records from a given domain

Example:

D("example.com!external", REGISTRAR, DnsProvider(R53),
  A("test", "8.8.8.8")
);

D("example.com!internal", REGISTRAR, DnsProvider(R53),
  INCLUDE("example.com!external"),
  A("home", "127.0.0.1")
);

MX( name, priority, target, modifiers... )

MX adds an MX record to the domain.

Priority should be a number.

Target should be a string representing the MX target. If it is a single label we will assume it is a relative name on the current domain. If it contains any dots, it should be a fully qualified domain name, ending with a ..

Example:

D("example.com", REGISTRAR, DnsProvider(R53),
  MX("@", 5, "mail"), // mx example.com -> mail.example.com
  MX("sub", 10, "mail.foo.com.")
);

NAMESERVER( name, modifiers... )

NAMESERVER() instructs DNSControl to inform the domain’s registrar where to find this zone. For some registrars this will also add NS records to the zone itself.

This takes exactly one argument: the name of the nameserver. It must end with a “.” if it is a FQDN, just like all targets.

This is different than the NS() function, which inserts NS records in the current zone and accepts a label. NS() is useful for downward delegations. NAMESERVER() is for informing upstream delegations.

For more information, refer to this page.

Example:

D("example.com", REGISTRAR, .... ,
  DnsProvider(route53, 0),
  // Replace the nameservers:
  NAMESERVER("ns1.myserver.com."),
  NAMESERVER("ns2.myserver.com."),
);

D("example2.com", REGISTRAR, .... ,
  // Add these two additional nameservers to the existing list of nameservers.
  NAMESERVER("ns1.myserver.com."),
  NAMESERVER("ns2.myserver.com."),
);

The difference between NS() and NAMESERVER()

Nameservers are one of the least understood parts of DNS, so a little extra explanation is required.

  • NS() lets you add an NS record to a zone, just like A() adds an A record to the zone. This is generally used to delegate a subzone.

  • The NAMESERVER() directive speaks to the Registrar about how the parent should delegate the zone.

Since the parent zone could be completely unrelated to the current zone, changes made by NAMESERVER() have to be done by an API call to the registrar, who then figures out what to do. For example, if I use NAMESERVER() in the zone stackoverflow.com, DNSControl talks to the registrar who does the hard work of talking to the people that control .com. If the domain was gmeet.io, the registrar does the right thing to talk to the people that control .io.

(A better name might have been PARENTNAMESERVER() but we didn’t think of that at the time.)

Each registrar handles delegations differently. Most use the NAMESERVER() targets to update the delegation, adding NS records to the parent zone as required. Some providers restrict the names to hosts they control. Others may require you to add the NS records to the parent domain manually.

How to not change the parent NS records?

If dnsconfig.js has zero NAMESERVER() commands for a domain, it will use the API to remove all non-default nameservers.

If dnsconfig.js has 1 or more NAMESERVER() commands for a domain, it will use the API to add those nameservers (unless, of course, they already exist).

So how do you tell DNSControl not to make any changes at all? Use the special Registrar called “NONE”. It makes no changes.

It looks like this:

var REG_THIRDPARTY = NewRegistrar('ThirdParty', 'NONE')
D("mydomain.com", REG_THIRDPARTY,
  ...
)

NAMESERVER_TTL( ttl )

TTL sets the TTL on the domain apex NS RRs defined by NAMESERVER.

The value can be an integer or a string. See TTL for examples.

Example:

D('example.com', REGISTRAR, DnsProvider('R53'),
  NAMESERVER_TTL('2d'),
  NAMESERVER('ns')
);

NO_PURGE( )

NO_PURGE indicates that records should not be deleted from a domain. Records will be added and updated, but not removed.

NO_PURGE is generally used in very specific situations:

  • A domain is managed by some other system and DNSControl is only used to insert a few specific records and/or keep them updated. For example a DNS Zone that is managed by Active Directory, but DNSControl is used to update a few, specific, DNS records. In this case we want to specify the DNS records we are concerned with but not delete all the other records. This is a risky use of NO_PURGE since, if NO_PURGE was removed (or buggy) there is a chance you could delete all the other records in the zone, which could be a disaster. That said, domains with some records updated using Dynamic DNS have no other choice.
  • To work-around a pseudo record type that is not supported by DNSControl. For example some providers have a fake DNS record type called “URL” which creates a redirect. DNSControl normally deletes these records because it doesn’t understand them. NO_PURGE will leave those records alone.

In this example DNSControl will insert “foo.example.com” into the zone, but otherwise leave the zone alone. Changes to “foo”’s IP address will update the record. Removing the A(“foo”, …) record from dnscontrol will leave the record in place.

Example:

D("example.com", .... , NO_PURGE,
  A("foo","1.2.3.4")
);

The main caveat of NO_PURGE is that intentionally deleting records becomes more difficult. Suppose a NO_PURGE zone has an record such as A(“ken”, “1.2.3.4”). Removing the record from dnsconfig.js will not delete “ken” from the domain. DNSControl has no way of knowing the record was deleted from the file The DNS record must be removed manually. Users of NO_PURGE are prone to finding themselves with an accumulation of orphaned DNS records. That’s easy to fix for a small zone but can be a big mess for large zones.

Not all providers support NO_PURGE. For example the BIND provider rewrites zone files from scratch each time, which precludes supporting NO_PURGE. DNSControl will exit with an error if NO_PURGE is used on a driver that does not support it.

There is also PURGE command for completeness. PURGE is the default, thus this command is a no-op.

NS( name, target, modifiers... )

NS adds a NS record to the domain. The name should be the relative label for the domain.

The name may not be @ (the bare domain), as that is controlled via NAMESERVER(). The difference between NS() and NAMESERVER() is explained in the NAMESERVER() description.

Target should be a string representing the NS target. If it is a single label we will assume it is a relative name on the current domain. If it contains any dots, it should be a fully qualified domain name, ending with a ..

Example:

D("example.com", REGISTRAR, DnsProvider("R53"),
  NS("foo", "ns1.example2.com."), // Delegate ".foo.example.com" zone to another server.
  NS("foo", "ns2.example2.com."), // Delegate ".foo.example.com" zone to another server.
  A("ns1.example2.com", "10.10.10.10"), // Glue records
  A("ns2.example2.com", "10.10.10.20"), // Glue records
);

PTR( name, target, modifiers... )

PTR adds a PTR record to the domain.

The name is normally a the relative label for the domain, or a FQDN that ends with .. If magic mode is enabled (see below) it can also be an IP address, which will be replaced by the proper string automatically, thus saving the user from having to reverse the IP address manually.

Target should be a string representing the FQDN of a host. Like all FQDNs in DNSControl, it must end with a ..

Magic Mode:

PTR records are complex and typos are common. Therefore DNSControl enables features to save labor and prevent typos. This magic is only enabled when the domain ends with in-addr.arpa. or ipv6.arpa..

Automatic IP-to-reverse: If the name is a valid IP address, DNSControl will replace it with a string that is appropriate for the domain. That is, if the domain ends with in-addr.arpa (no .) and name is a valid IPv4 address, the name will be replaced with the correct string to make a reverse lookup for that address. IPv6 is properly handled too.

Extra Validation: DNSControl considers it an error to include a name that is inappropriate for the domain. For example PTR('1.2.3.4', 'f.co.') is valid for the domain D("3.2.1.in-addr.arpa', but DNSControl will generate an error if the domain is D("9.9.9.in-addr.arpa',. This is because 1.2.3.4 is contained in 1.2.3.0/24 but not 9.9.9.0/24. This validation works for IPv6, IPv4, and RFC2317 “Classless in-addr.arpa delegation” domains.

Automatic truncation: DNSControl will automatically truncate FQDNs as needed. If the name is a FQDN ending with ., DNSControl will verify that the name is contained within the CIDR block implied by domain. For example if name is 4.3.2.1.in-addr.arpa. (note the trailing .) and the domain is 2.1.in-addr.arpa (no trailing .) then the name will be replaced with 4.3. Note that the output of REV('1.2.3.4') is 4.3.2.1.in-addr.arpa., which means the following are all equivalent:

  • PTR(REV('1.2.3.4'),
  • PTR('4.3.2.1.in-addr.arpa.'),
  • PTR('4.3', // Assuming the domain is 2.1.in-addr.arpa

All magic is RFC2317-aware. We use the first format listed in the RFC for both REV() and PTR(). The format is FIRST/MASK.C.B.A.in-addr.arpa where FIRST is the first IP address of the zone, MASK is the netmask of the zone (25-31 inclusive), and A, B, C are the first 3 octets of the IP address. For example 172.20.18.130/27 is located in a zone named 128/27.18.20.172.in-addr.arpa

Example:

D(REV('1.2.3.0/24'), REGISTRAR, DnsProvider(BIND),
  PTR('1', 'foo.example.com.'),
  PTR('2', 'bar.example.com.'),
  PTR('3', 'baz.example.com.'),
  // If the first parameter is a valid IP address, DNSControl will generate the correct name:
  PTR('1.2.3.10', 'ten.example.com.'),    // '10'
);

D(REV('9.9.9.128/25'), REGISTRAR, DnsProvider(BIND),
  PTR('9.9.9.129', 'first.example.com.'),
);

D(REV('2001:db8:302::/48'), REGISTRAR, DnsProvider(BIND),
  PTR('1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0', 'foo.example.com.'),  // 2001:db8:302::1
  // If the first parameter is a valid IP address, DNSControl will generate the correct name:
  PTR('2001:db8:302::2', 'two.example.com.'),                          // '2.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0'
  PTR('2001:db8:302::3', 'three.example.com.'),                        // '3.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0'
);

In the future we plan on adding a flag to A() which will insert the correct PTR() record if the appropriate .arpa domain has been defined.

PURGE( )

PURGE is the default setting for all domains. Therefore PURGE is a no-op. It is included for completeness only.

A domain with a mixture of NO_PURGE and PURGE parameters will abide by the last one.

These three examples all are equivalent.

PURGE is the default:

Example:

D("example.com", .... ,
);

Purge is the default, but we set it anyway:

Example:

D("example.com", .... ,
  PURGE,
);

Since the “last command wins”, this is the same as PURGE:

Example:

D("example.com", .... ,
  PURGE,
  NO_PURGE,
  PURGE,
  NO_PURGE,
  PURGE,
);

R53_ALIAS( name, target, ZONE_ID modifier )

R53_ALIAS is a Route53 specific virtual record type that points a record at either another record or an AWS entity (like a Cloudfront distribution, an ELB, etc…). It is analogous to a CNAME, but is usually resolved at request-time and served as an A record. Unlike CNAMEs, ALIAS records can be used at the zone apex (@)

Unlike the regular ALIAS directive, R53_ALIAS is only supported on Route53. Attempting to use R53_ALIAS on another provider than Route53 will result in an error.

The name should be the relative label for the domain.

Target should be a string representing the target. If it is a single label we will assume it is a relative name on the current domain. If it contains any dots, it should be a fully qualified domain name, ending with a ..

The Target can be any of:

  • CloudFront distribution: in this case specify the domain name that CloudFront assigned when you created your distribution (note that your CloudFront distribution must include an alternate domain name that matches the record you’re adding)
  • Elastic Beanstalk environment: specify the CNAME attribute for the environment. The environment must have a regionalized domain name. To get the CNAME, you can use either the AWS Console, AWS Elastic Beanstalk API, or the AWS CLI.
  • ELB load balancer: specify the DNS name that is associated with the load balancer. To get the DNS name you can use either the AWS Console (on the EC2 page, choose Load Balancers, select the right one, choose the description tab), ELB API, the AWS ELB CLI, or the AWS ELBv2 CLI.
  • S3 bucket (configured as website): specify the domain name of the Amazon S3 website endpoint in which you configured the bucket (for instance s3-website-us-east-2.amazonaws.com). For the available values refer to the Amazon S3 Website Endpoints.
  • Another Route53 record: specify the value of the name of another record in the same hosted zone.

For all the target type, excluding ‘another record’, you have to specify the Zone ID of the target. This is done by using the R53_ZONE record modifier.

The zone id can be found depending on the target type:

  • CloudFront distribution: specify Z2FDTNDATAQYW2
  • Elastic Beanstalk environment: specify the hosted zone ID for the region in which the environment has been created. Refer to the List of regions and hosted Zone IDs.
  • ELB load balancer: specify the value of the hosted zone ID for the load balancer. You can find it in the List of regions and hosted Zone IDs
  • S3 bucket (configured as website): specify the hosted zone ID for the region that you created the bucket in. You can find it in the List of regions and hosted Zone IDs
  • Another Route 53 record: you can either specify the correct zone id or do not specify anything and dnscontrol will figure out the right zone id. (Note: Route53 alias can’t reference a record in a different zone).

Example:

D('example.com', REGISTRAR, DnsProvider('ROUTE53'),
  R53_ALIAS('foo', 'A', 'bar'),                              // record in same zone
  R53_ALIAS('foo', 'A', 'bar', R53_ZONE('Z35SXDOTRQ7X7K')),  // record in same zone, zone specified
  R53_ALIAS('foo', 'A', 'blahblah.elasticloadbalancing.us-west-1.amazonaws.com.', R53_ZONE('Z368ELLRRE2KJ0')),     // a classic ELB in us-west-1
  R53_ALIAS('foo', 'A', 'blahblah.elasticbeanstalk.us-west-2.amazonaws.com.', R53_ZONE('Z38NKT9BP95V3O')),     // an Elastic Beanstalk environment in us-west-2
  R53_ALIAS('foo', 'A', 'blahblah-bucket.s3-website-us-west-1.amazonaws.com.', R53_ZONE('Z2F56UZL2M1ACD')),     // a website S3 Bucket in us-west-1
);

SOA( name, ns, mbox, refresh, retry, expire, minttl, modifiers... )

SOA adds an SOA record to a domain. The name should be @. ns and mbox are strings. The other fields are unsigned 32-bit ints.

Example:

D("example.com", REG_THIRDPARTY, DnsProvider("DNS_BIND"),
  SOA("@", "ns3.example.org.", "hostmaster.example.org.", 3600, 600, 604800, 1440),
);

Notes:

  • The serial number is managed automatically. It isn’t even a field in SOA().
  • Most providers automatically generate SOA records. They will ignore any SOA() statements.

There is more info about SOA in the documentation for the BIND provider.

SRV( name, priority, weight, port, target, modifiers... )

SRV adds a SRV record to a domain. The name should be the relative label for the record.

Priority, weight, and port are ints.

Example:

D("example.com", REGISTRAR, DnsProvider("GCLOUD"),
  // Create SRV records for a a SIP service:
  //               pr  w   port, target
  SRV('_sip._tcp', 10, 60, 5060, 'bigbox.example.tld.'),
  SRV('_sip._tcp', 10, 20, 5060, 'smallbox1.example.tld.'),
);

SSHFP( name, algorithm, type, value, modifiers... )

SSHFP contains a fingerprint of a SSH server which can be validated before SSH clients are establishing the connection.

Algorithm (type of the key) | ID | Algorithm | |—-|———–| | 0 | reserved | | 1 | RSA | | 2 | DSA | | 3 | ECDSA | | 4 | ED25519 |

Type (fingerprint format) | ID | Algorithm | |—-|———–| | 0 | reserved | | 1 | SHA-1 | | 2 | SHA-256 |

value is the fingerprint as a string.

Example:

SSHFP('@', 1, 1, '00yourAmazingFingerprint00'),

TLSA( name, usage, selector, type, certificate, modifiers... )

TLSA adds a TLSA record to a domain. The name should be the relative label for the record.

Usage, selector, and type are ints.

Certificate is a hex string.

Example:

D("example.com", REGISTRAR, DnsProvider("GCLOUD"),
  // Create TLSA record for certificate used on TCP port 443
  TLSA("_443._tcp", 3, 1, 1, "abcdef0"),
);

TXT( name, contents, modifiers... )

TXT adds an TXT record To a domain. The name should be the relative label for the record. Use @ for the domain apex.

The contents is either a single or multiple strings. To specify multiple strings, specify them as an array.

Each string is a JavaScript string (quoted using single or double quotes). The (somewhat complex) quoting rules of the DNS protocol will be done for you.

Modifiers can be any number of record modifiers or json objects, which will be merged into the record’s metadata.

Example:

    D("example.com", REGISTRAR, ....,
      TXT('@', '598611146-3338560'),
      TXT('listserve', 'google-site-verification=12345'),
      TXT('multiple', ['one', 'two', 'three']),  // Multiple strings
      TXT('quoted', 'any "quotes" and escapes? ugh; no worries!'),
      TXT('_domainkey', 't=y; o=-;'), // Escapes are done for you automatically.
      TXT('long', 'X'.repeat(300)) // Long strings are split automatically.
    );

NOTE: In the past, long strings had to be annotated with the keyword AUTOSPLIT. This is no longer required. The keyword is now a no-op.

Long strings

Strings that are longer than 255 octets (bytes) will be quietly split into 255-octets chunks or the provider may report an error if it does not handle multiple strings.

TXT record edge cases.

Most providers do not support the full possibilities of what a TXT record can store. DNSControl can not handle all the edge cases and incompatibles that providers have introduced. Instead, it stores the string(s) that you provide and passes them to the provider verbatim. The provider may opt to accept the data, fix it, or reject it. This happens early in the processing, long before the DNSControl talks to the provider’s API.

The RFCs specify that a TXT record stores one or more strings, each is up to 255 octets (bytes) long. We call these individual strings chunks. Each chunk may be zero to 255 octets long. There is no limit to the number of chunks in a TXT record, other than IP packet length restrictions. The contents of each chunk may be octets of value from 0x00 to 0xff.

In reality DNS Service Providers (DSPs) place many restrictions on TXT records.

Some DSPs only support a single string of 255 octets or fewer. Multiple strings, or any one string being longer than 255 octets will result in an error. One provider limits the string to 254 octets, which makes me think they’re code has an off-by-one error.

Some DSPs only support one string, but it may be of any length. Behind the scenes the provider splits it into 255-octet chunks (except the last one, of course).

Some DSPs support multiple strings, but API requests must be 512-bytes or fewer, and with quoting, escaping, and other encoding mishegoss you can’t be sure what will be permitted until you actually try it.

Regardless of the quantity and length of strings, some providers ban double quotes, back-ticks, or other chars.

Testing the support of a provider

How can you tell if a provider will support a particular TXT() record?

Include the TXT() record in a D() as usual, along with the DnsProvider() for that provider. Run dnscontrol check to see if any errors are produced. The check command does not talk to the provider’s API, thus permitting you to do this without having an account at that provider.

What if the provider rejects a string that is supported?

Suppose I can create the TXT record using the DSP’s web portal but DNSControl rejects the string?

It is possible that the provider code in DNSControl rejects strings that the DSP accepts. This is because the test is done in code, not by querying the provider’s API. It is possible that the code was written to work around a bug (such as rejecting a string with a back-tick) but now that bug has been fixed.

All such checks are in providers/${providername}/auditrecords.go. You can try removing the check that you feel is in error and see if the provider’s API accepts the record. You can do this by running the integration tests, or by simply adding that record to an existing dnsconfig.js and seeing if dnscontrol push is able to push that record into production. (Be careful if you are testing this on a domain used in production.)

URL( name, modifiers... )

Documentation needed.

URL301( name, modifiers... )

Documentation needed.

Record Modifiers

( )

DMARC Builder

dnscontrol contains a DMARC_BUILDER which can be used to simply create DMARC policies for your domains.

Example

Simple example

DMARC_BUILDER({
  policy: 'reject',
  ruf: [
    'mailto:mailauth-reports@example.com',
  ],
})

This yield the following record:

@   IN  TXT "v=DMARC1; p=reject; ruf=mailto:mailauth-reports@example.com"

Advanced example

DMARC_BUILDER({
  policy: 'reject',
  subdomainPolicy: 'quarantine',
  percent: 50,
  alignmentSPF: 'r',
  alignmentDKIM: 'strict',
  rua: [
    'mailto:mailauth-reports@example.com',
    'https://dmarc.example.com/submit',
  ],
  ruf: [
    'mailto:mailauth-reports@example.com',
  ],
  failureOptions: '1',
  reportInterval: '1h',
}),

DMARC_BUILDER({
  label: 'insecure',
  policy: 'none',
  ruf: [
    'mailto:mailauth-reports@example.com',
  ],
  failureOptions: {
      SPF: false,
      DKIM: true,
  },
})

This yields the following records:

@           IN  TXT "v=DMARC1; p=reject; sp=quarantine; adkim=s; aspf=r; pct=50; rua=mailto:mailauth-reports@example.com,https://dmarc.example.com/submit; ruf=mailto:mailauth-reports@example.com; fo=1; ri=3600"
insecure    IN  TXT "v=DMARC1; p=none; ruf=mailto:mailauth-reports@example.com; fo=d"

Parameters

  • label: The DNS label for the DMARC record (_dmarc prefix is added, default: '@')
  • policy: The DMARC policy (p=), must be one of 'none', 'quarantine', 'reject'
  • subdomainPolicy: The DMARC policy for subdomains (sp=), must be one of 'none', 'quarantine', 'reject' (optional)
  • alignmentSPF: 'strict'/'s' or 'relaxed'/'r' alignment for SPF (aspf=, default: 'r')
  • alignmentDKIM: 'strict'/'s' or 'relaxed'/'r' alignment for DKIM (adkim=, default: 'r')
  • percent: Number between 0 and 100, percentage for which policies are applied (pct=, default: 100)
  • rua: Array of aggregate report targets (optional)
  • ruf: Array of failure report targets (optional)
  • failureOptions: Object or string; Object containing booleans SPF and DKIM, string is passed raw (fo=, default: '0')
  • failureFormat: Format in which failure reports are requested (rf=, default: 'afrf')
  • reportInterval: Interval in which reports are requested (ri=)
  • ttl: Input for TTL method (optional)

Caveats

  • TXT records are automatically split using AUTOSPLIT.
  • URIs in the rua and ruf arrays are passed raw. You must percent-encode all commas and exclamation points in the URI itself.

R53_ZONE( zone_id )

R53_ZONE sets the required Route53 hosted zone id in a R53_ALIAS record.

This directive has no impact when used in anything else than a R53_ALIAS.

Please refer to the R53_ALIAS directive for usage.

TTL( ttl )

TTL sets the TTL for a single record only. This will take precedence over the domain’s DefaultTTL if supplied.

The value can be:

  • An integer (number of seconds). Example: 600
  • A string: Integer with single-letter unit: Example: 5m
  • The unit denotes:
    • s (seconds)
    • m (minutes)
    • h (hours)
    • d (days)
    • w (weeks)
    • n (nonths) (30 days in a nonth)
    • y (years) (If you set a TTL to a year, we assume you also do crossword puzzles in pen. Show off!)
    • If no unit is specified, the default is seconds.
  • We highly recommend using units instead of the number of seconds. Would your coworkers understand your intention better if you wrote 14400 or '4h'?

Example:

D('example.com', REGISTRAR, DnsProvider('R53'),
  DefaultTTL(2000),
  A('@','1.2.3.4'), // uses default
  A('foo', '2.3.4.5', TTL(500)), // overrides default
  A('demo1', '3.4.5.11', TTL('5d')),  // 5 days
  A('demo2', '3.4.5.12', TTL('5w')),  // 5 weeks
);