The Ingress Controller can configure Kong specific features using several Custom Resource Definitions(CRDs).
Following CRDs enables users to declaratively configure all aspects of Kong:
- KongPlugin: This resource corresponds to the Plugin entity in Kong.
- KongIngress: This resource provides fine-grained control over all aspects of proxy behaviour like routing, load-balancing, and health checking. It serves as an "extension" to the Ingress resources in Kubernetes.
- KongConsumer: This resource maps to the Consumer entity in Kong.
- TCPIngress: This resource can configure TCP-based routing in Kong for non-HTTP services running inside Kubernetes.
- KongCredential (Deprecated): This resource maps to a credential (key-auth, basic-auth, jwt, hmac-auth) that is associated with a specific KongConsumer.
This resource provides an API to configure plugins inside Kong using Kubernetes-style resources.
Please see the concept document for how the resource should be used.
The following snippet shows the properties available in KongPlugin resource:
apiVersion: configuration.konghq.com/v1
kind: KongPlugin
metadata:
name: <object name>
namespace: <object namespace>
labels:
global: "true" # optional, if set, then the plugin will be executed
# for every request that Kong proxies
# please note the quotes around true
disabled: <boolean> # optionally disable the plugin in Kong
config: # configuration for the plugin
key: value
plugin: <name-of-plugin> # like key-auth, rate-limiting etc
config
contains a list ofkey
andvalue
required to configure the plugin. All configuration values specific to the type of plugin go in here. Please read the documentation of the plugin being configured to set values in here. For any plugin in Kong, anything that goes in theconfig
JSON key in the Admin API request, goes into theconfig
YAML key in this resource. Please use a valid JSON to YAML convertor and place the content under theconfig
key in the YAML above.plugin
field determines the name of the plugin in Kong. This field was introduced in Kong Ingress Controller 0.2.0.- Setting a label
global
to"true"
will result in the plugin being applied globally in Kong, meaning it will be executed for every request that is proxied via Kong.
Please note: validation of the configuration fields is left to the user by default. It is advised to setup and use the admission validating controller to catch user errors.
The plugins can be associated with Ingress
or Service object in Kubernetes using plugins.konghq.com
annotation.
Example:
Given the following plugin:
apiVersion: configuration.konghq.com/v1
kind: KongPlugin
metadata:
name: request-id
config:
header_name: my-request-id
plugin: correlation-id
It can be applied to a service by annotating like:
apiVersion: v1
kind: Service
metadata:
name: myapp-service
labels:
app: myapp-service
annotations:
plugins.konghq.com: request-id
spec:
ports:
- port: 80
targetPort: 80
protocol: TCP
name: myapp-service
selector:
app: myapp-service
It can be applied to a specific ingress (route or routes):
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
name: demo-example-com
annotations:
plugins.konghq.com: request-id
spec:
rules:
- host: example.com
http:
paths:
- path: /bar
backend:
serviceName: echo
servicePort: 80
A plugin can also be applied to a specific KongConsumer by adding
plugins.konghq.com
annotation to the KongConsumer resource.
Please follow the Using the KongPlugin resource guide for details on how to use this resource.
A KongClusterPlugin
is same as KongPlugin
resource. The only difference
being that it is a Kubernetes cluster-level resource instead of a
namespaced resource.
Please consult the KongPlugin section for details.
Example:
KongClusterPlugin example:
apiVersion: configuration.konghq.com/v1
kind: KongClusterPlugin
metadata:
name: request-id
config:
header_name: my-request-id
plugin: correlation-id
Ingress resource spec in Kubernetes can define routing policies
based on HTTP Host header and paths.
While this is sufficient in most cases,
sometimes, users may want more control over routing at the Ingress level.
KongIngress
serves as an "extension" to Ingress resource.
It is not meant as a replacement to the
Ingress
resource in Kubernetes.
Please read the concept document for why this resource exists and how it relates to the existing Ingress resource.
Using KongIngress
, all properties of Upstream,
Service and
Route entities in Kong related to an Ingress resource
can be modified.
Once a KongIngress
resource is created, it needs to be associated with
an Ingress or Service resource using the following annotation:
configuration.konghq.com: kong-ingress-resource-name
Specifically,
- To override any properties related to health-checking, load-balancing, or details specific to a service, add the annotation to the Kubernetes Service that is being exposed via the Ingress API.
- To override routing configuration (like protocol or method based routing), add the annotation to the Ingress resource.
Please follow the Using the KongIngress resource guide for details on how to use this resource.
For reference, the following is a complete spec for KongIngress:
apiVersion: configuration.konghq.com/v1
kind: KongIngress
metadata:
name: configuration-demo
upstream:
slots: 10
hash_on: none
hash_fallback: none
healthchecks:
threshold: 25
active:
concurrency: 10
healthy:
http_statuses:
- 200
- 302
interval: 0
successes: 0
http_path: "/"
timeout: 1
unhealthy:
http_failures: 0
http_statuses:
- 429
interval: 0
tcp_failures: 0
timeouts: 0
passive:
healthy:
http_statuses:
- 200
successes: 0
unhealthy:
http_failures: 0
http_statuses:
- 429
- 503
tcp_failures: 0
timeouts: 0
proxy:
protocol: http
path: /
connect_timeout: 10000
retries: 10
read_timeout: 10000
write_timeout: 10000
route:
methods:
- POST
- GET
regex_priority: 0
strip_path: false
preserve_host: true
protocols:
- http
- https
The Ingress resource in Kubernetes is HTTP-only. This custom resource is modeled similar to the Ingress resource but for TCP and TLS SNI based routing purposes:
apiVersion: configuration.konghq.com/v1beta1
kind: TCPIngress
metadata:
name: <object name>
namespace: <object namespace>
spec:
rules:
- host: <SNI, optional>
port: <port on which to expose this service, required>
backend:
serviceName: <name of the kubernetes service, required>
servicePort: <port number to forward on the service, required>
If host
is not specified, then port-based TCP routing is performed. Kong
doesn't care about the content of TCP stream in this case.
If host
is specified, then Kong expects the TCP stream to be TLS-encrypted
and Kong will terminate the TLS session based on the SNI.
Also note that, the port in this case should be configured with ssl
parameter
in Kong.
This custom resource configures a consumer in Kong:
The following snippet shows the field available in the resource:
apiVersion: configuration.konghq.com/v1
kind: KongConsumer
metadata:
name: <object name>
namespace: <object namespace>
username: <user name>
custom_id: <custom ID>
An example:
apiVersion: configuration.konghq.com/v1
kind: KongConsumer
metadata:
name: consumer-team-x
username: team-X
When this resource is created, a corresponding consumer entity will be created in Kong.
This custom resource can be used to configure a consumer specific
entities in Kong.
The resource reference the KongConsumer resource via the consumerRef
key.
The validation of the config object is left up to the user.
apiVersion: configuration.konghq.com/v1
kind: KongCredential
metadata:
name: credential-team-x
consumerRef: consumer-team-x
type: key-auth
config:
key: 62eb165c070a41d5c1b58d9d3d725ca1
The following credential types can be provisioned using the KongCredential resource:
key-auth
for Key authenticationbasic-auth
for Basic authenticaitonhmac-auth
for HMAC authenticationjwt
for JWT based authenticationoauth2
for Oauth2 Client credentialsacl
for ACL group associations
Please ensure that all fields related to the credential in Kong
are present in the definition of KongCredential's config
section.
Please refer to the using the Kong Consumer and Credential resource guide for details on how to use this resource.