The main method that gets called is ServeDNS
. It has three parameters:
- a
context.Context
; dns.ResponseWriter
that is, basically, the client's connection;*dns.Msg
the request from the client.
ServeDNS
returns two values, a response code and an error. If the error is not nil, CoreDNS
will return a SERVFAIL to the client. The response code tells CoreDNS if a reply has been
written by the plugin chain or not. In the latter case CoreDNS will take care of that.
CoreDNS treats:
- SERVFAIL (dns.RcodeServerFailure)
- REFUSED (dns.RcodeRefused)
- FORMERR (dns.RcodeFormatError)
- NOTIMP (dns.RcodeNotImplemented)
as special and will then assume nothing has been written to the client. In all other cases it assumes something has been written to the client (by the plugin).
The example plugin shows a bare-bones implementation that can be used as a starting point for your plugin. This plugin has tests and extensive comments in the code.
See a couple of blog posts on how to write and add plugin to CoreDNS:
- https://blog.coredns.io/2017/03/01/how-to-add-plugins-to-coredns/
- https://blog.coredns.io/2016/12/19/writing-plugins-for-coredns/, slightly older, but useful.
If your plugin needs to output a log line you should use the plugin/pkg/log
package. This package
implements log levels. The standard way of outputting is: log.Info
for info level messages. The
levels available are log.Info
, log.Warning
, log.Error
, log.Debug
. Each of these also has
a f
variant. The plugin's name should be included, by using the log package like so:
import clog "github.com/coredns/coredns/plugin/pkg/log"
var log = clog.NewWithPlugin("whoami")
log.Info("message") // outputs: [INFO] plugin/whoami: message
In general, logging should be left to the higher layers by returning an error. However, if there is
a reason to consume the error and notify the user, then logging in the plugin itself can be
acceptable. The Debug*
functions only output something when the debug plugin is loaded in the
server.
When exporting metrics the Namespace should be plugin.Namespace
(="coredns"), and the
Subsystem must be the name of the plugin. The README.md for the plugin should then also contain
a Metrics section detailing the metrics.
If the plugin supports signalling readiness it should have a Ready section detailing how it
works, and implement the ready.Readiness
interface.
See the plugin/pkg/reuseport for Listen
and ListenPacket
functions. Using these functions makes
your plugin handle reload events better.
Every request get a context.Context these are pre-filled with 2 values:
Key
: holds a pointer to the current server, this can be useful for logging or metrics. It is infact used in the metrics plugin to tie a request to a specific (internal) server.LoopKey
: holds an integer to detect loops within the current context. The file plugin uses this to detect loops when resolving CNAMEs.
Each plugin should have a README.md explaining what the plugin does and how it is configured. The file should have the following layout:
- Title: use the plugin's name
- Subsection titled: "Name" with PLUGIN - one line description.
- Subsection titled: "Description" has a longer description.
- Subsection titled: "Syntax", syntax and supported directives.
- Subsection titled: "Examples"
More sections are of course possible.
We use the Unix manual page style:
- The name of plugin in the running text should be italic: plugin.
- all CAPITAL: user supplied argument, in the running text references this use strong text:
**
: EXAMPLE. - Optional text: in block quotes:
[optional]
. - Use three dots to indicate multiple options are allowed:
arg...
. - Item used literal:
literal
.
Please be sure to use example.org
or example.net
in any examples and tests you provide. These
are the standard domain names created for this purpose.
In a perfect world the following would be true for plugin: "Either you are responsible for a zone or not". If the answer is "not", the plugin should call the next plugin in the chain. If "yes" it should handle all names that fall in this zone and the names below - i.e. it should handle the entire domain and all subdomains.
. {
file example.org db.example
}
In this example the file plugin is handling all names below (and including) example.org
. If
a query comes in that is not a subdomain (or equal to) example.org
the next plugin is called.
Now, the world isn't perfect, and there are may be reasons to "fallthrough" to the next plugin, meaning a plugin is only responsible for a subset of names within the zone.
The fallthrough
directive should optionally accept a list of zones. Only queries for records
in one of those zones should be allowed to fallthrough. See plugin/pkg/fallthrough
for the
implementation.
Using a custom ResponseWriter
, a plugin can mutate a response when another plugin further down the chain writes the response to the client.
If a plugin mutates a response it MUST make a copy of the entire response before doing so. A
response is a pointer to a dns.Msg
and as such you will be manipulating the original response,
which could have been generated from a data store. E.g. the file plugin creates a response that
the rewrite plugin then rewrites; not copying the data, means it's also mutating the data of
the file's data store. A response can be copied by using the Copy()
method.
Some general guidelines:
- logging time duration should be done in seconds (call the
Seconds()
method on any duration). - keep logging to a minimum.
- call the main config parse function just
parse
. - try to minimize the number of knobs in the configuration.
- use
plugin.Error()
to wrap errors returned from thesetup
function.
Plugins for CoreDNS can live out-of-tree, plugin.cfg
defaults to CoreDNS' repo but other
repos work just as well. So when do we consider the inclusion of a new plugin in the main repo?
- First, the plugin should be useful for other people. "Useful" is a subjective term. We will probably need to further refine this.
- It should be sufficiently different from other plugin to warrant inclusion.
- Current internet standards need be supported: IPv4 and IPv6, so A and AAAA records should be handled (if your plugin is in the business of dealing with address records that is).
- It must have tests.
- It must have a README.md for documentation.