Contour Logo

Blog

Jan 1, 0001

HTTPProxy Reference

The Ingress object was added to Kubernetes in version 1.1 to describe properties of a cluster-wide reverse HTTP proxy. Since that time, the Ingress object has not progressed beyond the beta stage, and its stagnation inspired an explosion of annotations to express missing properties of HTTP routing.

The goal of the HTTPProxy (previously IngressRoute) Custom Resource Definition (CRD) is to expand upon the functionality of the Ingress API to allow for a richer user experience as well addressing the limitations of the latter’s use in multi tenent environments.

Key HTTPProxy Benefits

  • Safely supports multi-team Kubernetes clusters, with the ability to limit which Namespaces may configure virtual hosts and TLS credentials.
  • Enables including of routing configuration for a path or domain from another HTTPProxy, possibly in another Namespace.
  • Accepts multiple services within a single route and load balances traffic across them.
  • Natively allows defining service weighting and load balancing strategy without annotations.
  • Validation of HTTPProxy objects at creation time and status reporting for post-creation validity.

Ingress to HTTPProxy

A minimal Ingress object might look like:

# ingress.yaml
apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  name: basic
spec:
  rules:
  - host: foo-basic.bar.com
    http:
      paths:
      - backend:
          serviceName: s1
          servicePort: 80

This Ingress object, named basic, will route incoming HTTP traffic with a Host: header for foo-basic.bar.com to a Service named s1 on port 80. Implementing similar behavior using an HTTPProxy looks like this:

{% highlight yaml linenos %}

httpproxy.yaml

apiVersion: projectcontour.io/v1 kind: HTTPProxy metadata: name: basic spec: virtualhost: fqdn: foo-basic.bar.com routes: - conditions: - prefix: / services: - name: s1 port: 80 {% endhighlight %}

Lines 1-5: As with all other Kubernetes objects, an HTTPProxy needs apiVersion, kind, and metadata fields.

Lines 7-8: The presence of the virtualhost field indicates that this is a root HTTPProxy that is the top level entry point for this domain. The fqdn field specifies the fully qualified domain name that will be used to match against Host: HTTP headers.

Lines 9-14: Each HTTPProxy must have one or more routes, each of which must have one or more services which will handle the HTTP traffic. In addition, each route may have one or more conditions to match against.

Lines 12-14: The services field is an array of named Service & Port combinations that will be used for this HTTPProxy path. HTTP traffic will be sent directly to the Endpoints corresponding to the Service.

Interacting with HTTPProxies

As with all Kubernetes objects, you can use kubectl to create, list, describe, edit, and delete HTTPProxy CRDs.

Creating an HTTPProxy:

$ kubectl create -f basic.httpproxy.yaml
httpproxy "basic" created

Listing HTTPProxies:

$ kubectl get httpproxy
NAME      AGE
basic     24s

Describing HTTPProxy:

$ kubectl describe httpproxy basic
Name:         basic
Namespace:    default
Labels:       <none>
API Version:  projectcontour.io/v1
Kind:         HTTPProxy
Metadata:
  Cluster Name:
  Creation Timestamp:  2019-07-05T19:26:54Z
  Resource Version:    19373717
  Self Link:           /apis/projectcontour.io/v1/namespaces/default/httpproxy/basic
  UID:                 6036a9d7-8089-11e8-ab00-f80f4182762e
Spec:
  Routes:
    Conditions:
      Prefix: /
    Services:
      Name:  s1
      Port:  80
  Virtualhost:
    Fqdn:  foo-basic.bar.com
Events:    <none>

Deleting HTTPProxies:

$ kubectl delete httpproxy basic
httpproxy "basic" deleted

HTTPProxy API Specification

There are a number of working examples of HTTPProxy objects in the examples/example-workload directory.

We will use these examples as a mechanism to describe HTTPProxy API functionality.

Virtual Host Configuration

Fully Qualified Domain Name

Similar to Ingress, HTTPProxy support name-based virtual hosting. Name-based virtual hosts use multiple host names with the same IP address.

foo.bar.com --|                 |-> foo.bar.com s1:80
              | 178.91.123.132  |
bar.foo.com --|                 |-> bar.foo.com s2:80

Unlike Ingress, HTTPProxy only support a single root domain per HTTPProxy object. As an example, this Ingress object:

# ingress-name.yaml
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: name-example
spec:
  rules:
  - host: foo1.bar.com
    http:
      paths:
      - backend:
          serviceName: s1
          servicePort: 80
  - host: bar1.bar.com
    http:
      paths:
      - backend:
          serviceName: s2
          servicePort: 80

must be represented by two different HTTPProxy objects:

# httpproxy-name.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: name-example-foo
  namespace: default
spec:
  virtualhost:
    fqdn: foo1.bar.com
  routes:
    - services:
      - name: s1
        port: 80
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: name-example-bar
  namespace: default
spec:
  virtualhost:
    fqdn: bar1.bar.com
  routes:
    - services:
        - name: s2
          port: 80

TLS

HTTPProxy follows a similar pattern to Ingress for configuring TLS credentials.

You can secure a HTTPProxy by specifying a Secret that contains TLS private key and certificate information. If multiple HTTPProxies utilize the same Secret, the certificate must include the necessary Subject Authority Name (SAN) for each fqdn.

Contour (via Envoy) requires that clients send the Server Name Indication (SNI) TLS extension so that requests can be routed to the correct virtual host. Virtual hosts are strongly bound to SNI names. This means that the Host header in HTTP requests must match the SNI name that was sent at the start of the TLS session.

Contour also follows a “secure first” approach. When TLS is enabled for a virtual host, any request to the insecure port is redirected to the secure interface with a 301 redirect. Specific routes can be configured to override this behavior and handle insecure requests by enabling the spec.routes.permitInsecure parameter on a Route.

The TLS secret must contain keys named tls.crt and tls.key that contain the certificate and private key to use for TLS, e.g.:

# ingress-tls.secret.yaml
apiVersion: v1
data:
  tls.crt: base64 encoded cert
  tls.key: base64 encoded key
kind: Secret
metadata:
  name: testsecret
  namespace: default
type: kubernetes.io/tls

The HTTPProxy can be configured to use this secret using tls.secretName property:

# httpproxy-tls.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: tls-example
  namespace: default
spec:
  virtualhost:
    fqdn: foo2.bar.com
    tls:
      secretName: testsecret
  routes:
    - services:
        - name: s1
          port: 80

If the tls.secretName property contains a slash, eg. somenamespace/somesecret then, subject to TLS Certificate Delegation, the TLS certificate will be read from somesecret in somenamespace. See TLS Certificate Delegation below for more information.

The TLS Minimum Protocol Version a vhost should negotiate can be specified by setting the spec.virtualhost.tls.minimumProtocolVersion:

  • 1.3
  • 1.2
  • 1.1 (Default)
Fallback Certificate

Contour provides virtual host based routing, so that any TLS request is routed to the appropriate service based on both the server name requested by the TLS client and the HOST header in the HTTP request.

As the HOST Header is encrypted during TLS handshake, it can’t be used for virtual host based routing unless the client sends HTTPS requests specifying hostname using the TLS server name, or the request is first decrypted using a default TLS certificate.

Some legacy TLS clients do not send the server name, so Envoy does not know how to select the right certificate. A fallback certificate is needed for these clients.

Note: The minimum TLS protocol version for any fallback request is defined by the minimum TLS protocol version set in the Contour configuration file. Enabling the fallback certificate is not compatible with TLS client authentication.

Configuration

First define the namespace/name in the [Contour configuration file][11] of a Kubernetes secret which will be used as the fallback certificate. Any HTTPProxy which enables fallback certificate delegation must have the fallback certificate delegated to the namespace in which the HTTPProxy object resides.

To do that, configure TLSCertificateDelegation to delegate the fallback certificate to specific or all namespaces (e.g. *) which should be allowed to enable the fallback certificate. Finally, for each root HTTPProxy, set the Spec.TLS.enableFallbackCertificate parameter to allow that HTTPProxy to opt-in to the fallback certificate routing.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: fallback-tls-example
  namespace: defaultub
spec:
  virtualhost:
    fqdn: fallback.bar.com
    tls:
      secretName: testsecret
      enableFallbackCertificate: true
  routes:
    - services:
        - name: s1
          port: 80
---
apiVersion: projectcontour.io/v1
kind: TLSCertificateDelegation
metadata:
  name: fallback-delegation
  namespace: www-admin
spec:
  delegations:
    - secretName: fallback-secret-name
      targetNamespaces:
      - "*"

Upstream TLS

A HTTPProxy can proxy to an upstream TLS connection by annotating the upstream Kubernetes Service or by specifying the upstream protocol in the HTTPProxy [services][10] field. Applying the projectcontour.io/upstream-protocol.tls annotation to a Service object tells Contour that TLS should be enabled and which port should be used for the TLS connection. The same configuration can be specified by setting the protocol name in the spec.routes.services[].protocol field on the HTTPProxy object. If both the annotation and the protocol field are specified, the protocol field takes precedence. By default, the upstream TLS server certificate will not be validated, but validation can be requested by setting the spec.routes.services[].validation field. This field has mandatory caSecret and subjectName fields, which specify the trusted root certificates with which to validate the server certificate and the expected server name.

Note: If spec.routes.services[].validation is present, spec.routes.services[].{name,port} must point to a Service with a matching projectcontour.io/upstream-protocol.tls Service annotation.

In the example below, the upstream service is named secure-backend and uses port 8443:

# httpproxy-example.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: example
spec:
  virtualhost:
    fqdn: www.example.com
  routes:
  - services:
    - name: secure-backend
      port: 8443
      validation:
        caSecret: my-certificate-authority
        subjectName: backend.example.com
# service-secure-backend.yaml
apiVersion: v1
kind: Service
metadata:
  name: secure-backend
  annotations:
    projectcontour.io/upstream-protocol.tls: "8443"
spec:
  ports:
  - name: https
    port: 8443
  selector:
    app: secure-backend
Error conditions

If the validation spec is defined on a service, but the secret which it references does not exist, Contour will reject the update and set the status of the HTTPProxy object accordingly. This helps prevent the case of proxying to an upstream where validation is requested, but not yet available.

Status:
  Current Status:  invalid
  Description:     route "/": service "tls-nginx": upstreamValidation requested but secret not found or misconfigured

TLS Certificate Delegation

In order to support wildcard certificates, TLS certificates for a *.somedomain.com, which are stored in a namespace controlled by the cluster administrator, Contour supports a facility known as TLS Certificate Delegation. This facility allows the owner of a TLS certificate to delegate, for the purposes of referencing the TLS certificate, permission to Contour to read the Secret object from another namespace.

The TLSCertificateDelegation resource defines a set of delegations in the spec. Each delegation references a secretName from the namespace where the TLSCertificateDelegation is created as well as describing a set of targetNamespaces in which the certificate can be referenced. If all namespaces should be able to reference the secret, then set "*" as the value of targetNamespaces (see example below).

apiVersion: projectcontour.io/v1
kind: TLSCertificateDelegation
metadata:
  name: example-com-wildcard
  namespace: www-admin
spec:
  delegations:
    - secretName: example-com-wildcard
      targetNamespaces:
      - example-com
    - secretName: another-com-wildcard
      targetNamespaces:
      - "*"
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: www
  namespace: example-com
spec:
  virtualhost:
    fqdn: foo2.bar.com
    tls:
      secretName: www-admin/example-com-wildcard
  routes:
    - services:
        - name: s1
          port: 80

In this example, the permission for Contour to reference the Secret example-com-wildcard in the admin namespace has been delegated to HTTPProxy objects in the example-com namespace. Also, the permission for Contour to reference the Secret another-com-wildcard from all namespaces has been delegated to all HTTPProxy objects in the cluster.

Conditions

Each Route entry in a HTTPProxy may contain one or more conditions. These conditions are combined with an AND operator on the route passed to Envoy.

Conditions can be either a prefix or a header condition.

Prefix conditions

For prefix, this adds a path prefix.

Up to one prefix condition may be present in any condition block.

Prefix conditions must start with a / if they are present.

Header conditions

For header conditions there is one required field, name, and five operator fields: present, contains, notcontains, exact, and notexact.

  • present is a boolean and checks that the header is present. The value will not be checked.

  • contains is a string, and checks that the header contains the string. notcontains similarly checks that the header does not contain the string.

  • exact is a string, and checks that the header exactly matches the whole string. notexact checks that the header does not exactly match the whole string.

Routes

HTTPProxy must have at least one route or include defined. Paths defined are matched using prefix conditions. In this example, any requests to multi-path.bar.com/blog or multi-path.bar.com/blog/* will be routed to the Service s2. All other requests to the host multi-path.bar.com will be routed to the Service s1.

# httpproxy-multiple-paths.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: multiple-paths
  namespace: default
spec:
  virtualhost:
    fqdn: multi-path.bar.com
  routes:
    - conditions:
      - prefix: / # matches everything else
      services:
        - name: s1
          port: 80
    - conditions:
      - prefix: /blog # matches `multi-path.bar.com/blog` or `multi-path.bar.com/blog/*`
      services:
        - name: s2
          port: 80

In the following example, we match on headers and send to different services, with a default route if those do not match.

# httpproxy-multiple-headers.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: multiple-paths
  namespace: default
spec:
  virtualhost:
    fqdn: multi-path.bar.com
  routes:
    - conditions:
      - header:
          name: x-os
          contains: ios
      services:
        - name: s1
          port: 80
    - conditions:
      - header:
          name: x-os
          contains: android
      services:
        - name: s2
          port: 80
    - services:
        - name: s3
          port: 80

Multiple Upstreams

One of the key HTTPProxy features is the ability to support multiple services for a given path:

# httpproxy-multiple-upstreams.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: multiple-upstreams
  namespace: default
spec:
  virtualhost:
    fqdn: multi.bar.com
  routes:
    - services:
        - name: s1
          port: 80
        - name: s2
          port: 80

In this example, requests for multi.bar.com/ will be load balanced across two Kubernetes Services, s1, and s2. This is helpful when you need to split traffic for a given URL across two different versions of an application.

Upstream Weighting

Building on multiple upstreams is the ability to define relative weights for upstream Services. This is commonly used for canary testing of new versions of an application when you want to send a small fraction of traffic to a specific Service.

# httpproxy-weight-shfiting.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: weight-shifting
  namespace: default
spec:
  virtualhost:
    fqdn: weights.bar.com
  routes:
    - services:
        - name: s1
          port: 80
          weight: 10
        - name: s2
          port: 80
          weight: 90

In this example, we are sending 10% of the traffic to Service s1, while Service s2 receives the remaining 90% of traffic.

HTTPProxy weighting follows some specific rules:

  • If no weights are specified for a given route, it’s assumed even distribution across the Services.
  • Weights are relative and do not need to add up to 100. If all weights for a route are specified, then the “total” weight is the sum of those specified. As an example, if weights are 20, 30, 20 for three upstreams, the total weight would be 70. In this example, a weight of 30 would receive approximately 42.9% of traffic (30/70 = .4285).
  • If some weights are specified but others are not, then it’s assumed that upstreams without weights have an implicit weight of zero, and thus will not receive traffic.

Request and Response Header Policies

Manipulating headers is also supported per-Service or per-Route. Headers can be set or removed from the request or response as follows:

per-Service:

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: header-manipulation
  namespace: default
spec:
  virtualhost:
    fqdn: headers.bar.com
  routes:
    - services:
        - name: s1
          port: 80
          requestHeadersPolicy:
            set:
              - name: X-Foo
                value: bar
            remove:
              - X-Baz
          responseHeadersPolicy:
            set:
              - name: X-Service-Name
                value: s1
            remove:
              - X-Internal-Secret

per-Route:

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: header-manipulation
  namespace: default
spec:
  virtualhost:
    fqdn: headers.bar.com
  routes:
    - services:
        - name: s1
          port: 80
      requestHeadersPolicy:
        set:
          - name: X-Foo
            value: bar
        remove:
          - X-Baz
      responseHeadersPolicy:
        set:
          - name: X-Service-Name
            value: s1
        remove:
          - X-Internal-Secret

In these examples we are setting the header X-Foo with value baz on requests and stripping X-Baz. We are then setting X-Service-Name on the response with value s1, and removing X-Internal-Secret.

Traffic mirroring

Per route a service can be nominated as a mirror. The mirror service will receive a copy of the read traffic sent to any non mirror service. The mirror traffic is considered read only, any response by the mirror will be discarded.

This service can be useful for recording traffic for later replay or for smoke testing new deployments.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: traffic-mirror
  namespace: default
spec:
  virtualhost:
    fqdn: www.example.com
  routes:
    - conditions:
      - prefix: /
      services:
        - name: www
          port: 80
        - name: www-mirror
          port: 80
          mirror: true

Response Timeout

Each Route can be configured to have a timeout policy and a retry policy as shown:

# httpproxy-response-timeout.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: response-timeout
  namespace: default
spec:
  virtualhost:
    fqdn: timeout.bar.com
  routes:
  - timeoutPolicy:
      response: 1s
      idle: 10s
    retryPolicy:
      count: 3
      perTryTimeout: 150ms
    services:
    - name: s1
      port: 80

In this example, requests to timeout.bar.com/ will have a response timeout policy of 1s. This refers to the time that spans between the point at which complete client request has been processed by the proxy, and when the response from the server has been completely processed.

  • timeoutPolicy.response This field can be any positive time period or “infinity”. The time period of 0s will also be treated as infinity. This timeout covers the time from the end of the client request to the end of the upstream response. By default, Envoy has a 15 second value for this timeout. More information can be found in Envoy’s documentation.
  • timeoutPolicy.idle This field can be any positive time period or “infinity”. The time period of 0s will also be treated as infinity. By default, there is no per-route idle timeout. Note that the default connection manager idle timeout of 5 minutes will apply if this is not set.

TimeoutPolicy durations are expressed as per the format specified in the ParseDuration documentation. Example input values: “300ms”, “5s”, “1m”. Valid time units are “ns”, “us” (or “µs”), “ms”, “s”, “m”, “h”. The string ‘infinity’ is also a valid input and specifies no timeout.

More information can be found in Envoy’s documentation

  • retryPolicy: A retry will be attempted if the server returns an error code in the 5xx range, or if the server takes more than retryPolicy.perTryTimeout to process a request.
    • retryPolicy.count specifies the maximum number of retries allowed. This parameter is optional and defaults to 1.
    • retryPolicy.perTryTimeout specifies the timeout per retry. If this field is greater than the request timeout, it is ignored. This parameter is optional. If left unspecified, timeoutPolicy.request will be used.

Load Balancing Strategy

Each route can have a load balancing strategy applied to determine which of its Endpoints is selected for the request. The following list are the options available to choose from:

  • RoundRobin: Each healthy upstream Endpoint is selected in round robin order (Default strategy if none selected).
  • WeightedLeastRequest: The least request strategy uses an O(1) algorithm which selects two random healthy Endpoints and picks the Endpoint which has fewer active requests. Note: This algorithm is simple and sufficient for load testing. It should not be used where true weighted least request behavior is desired.
  • Random: The random strategy selects a random healthy Endpoints.

More information on the load balancing strategy can be found in Envoy’s documentation.

The following example defines the strategy for the route / as WeightedLeastRequest.

# httpproxy-lb-strategy.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: lb-strategy
  namespace: default
spec:
  virtualhost:
    fqdn: strategy.bar.com
  routes:
    - conditions:
      - prefix: /
      services:
        - name: s1-strategy
          port: 80
        - name: s2-strategy
          port: 80
      loadBalancerPolicy:
        strategy: WeightedLeastRequest

Session Affinity

Session affinity, also known as sticky sessions, is a load balancing strategy whereby a sequence of requests from a single client are consistently routed to the same application backend. Contour supports session affinity on a per route basis with loadBalancerPolicy strategy: Cookie.

# httpproxy-sticky-sessions.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: httpbin
  namespace: default
spec:
  virtualhost:
    fqdn: httpbin.davecheney.com
  routes:
  - services:
    - name: httpbin
      port: 8080
    loadBalancerPolicy:
      strategy: Cookie
Limitations

Session affinity is based on the premise that the backend servers are robust, do not change ordering, or grow and shrink according to load. None of these properties are guaranteed by a Kubernetes cluster and will be visible to applications that rely heavily on session affinity.

Any perturbation in the set of pods backing a service risks redistributing backends around the hash ring.

Per route health checking

Active health checking can be configured on a per route basis. Contour supports HTTP health checking and can be configured with various settings to tune the behavior.

During HTTP health checking Envoy will send an HTTP request to the upstream Endpoints. It expects a 200 response if the host is healthy. The upstream host can return 503 if it wants to immediately notify Envoy to no longer forward traffic to it. It is important to note that these are health checks which Envoy implements and are separate from any other system such as those that exist in Kubernetes.

# httpproxy-health-checks.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: health-check
  namespace: default
spec:
  virtualhost:
    fqdn: health.bar.com
  routes:
  - conditions:
    - prefix: /
    healthCheckPolicy:
      path: /healthy
      intervalSeconds: 5
      timeoutSeconds: 2
      unhealthyThresholdCount: 3
      healthyThresholdCount: 5
    services:
      - name: s1-health
        port: 80
      - name: s2-health
        port: 80

Health check configuration parameters:

  • path: HTTP endpoint used to perform health checks on upstream service (e.g. /healthz). It expects a 200 response if the host is healthy. The upstream host can return 503 if it wants to immediately notify downstream hosts to no longer forward traffic to it.
  • host: The value of the host header in the HTTP health check request. If left empty (default value), the name “contour-envoy-healthcheck” will be used.
  • intervalSeconds: The interval (seconds) between health checks. Defaults to 5 seconds if not set.
  • timeoutSeconds: The time to wait (seconds) for a health check response. If the timeout is reached the health check attempt will be considered a failure. Defaults to 2 seconds if not set.
  • unhealthyThresholdCount: The number of unhealthy health checks required before a host is marked unhealthy. Note that for http health checking if a host responds with 503 this threshold is ignored and the host is considered unhealthy immediately. Defaults to 3 if not defined.
  • healthyThresholdCount: The number of healthy health checks required before a host is marked healthy. Note that during startup, only a single successful health check is required to mark a host healthy.

WebSocket Support

WebSocket support can be enabled on specific routes using the enableWebsockets field:

# httpproxy-websockets.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: chat
  namespace: default
spec:
  virtualhost:
    fqdn: chat.example.com
  routes:
    - services:
        - name: chat-app
          port: 80
    - conditions:
      - prefix: /websocket
      enableWebsockets: true # Setting this to true enables websocket for all paths that match /websocket
      services:
        - name: chat-app
          port: 80

Permit Insecure

A HTTPProxy can be configured to permit insecure requests to specific Routes. In this example, any request to foo2.bar.com/blog will not receive a 301 redirect to HTTPS, but the / route will:

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: tls-example-insecure
  namespace: default
spec:
  virtualhost:
    fqdn: foo2.bar.com
    tls:
      secretName: testsecret
  routes:
    - services:
        - name: s1
          port: 80
    - conditions:
      - prefix: /blog
      permitInsecure: true
      services:
        - name: s2
          port: 80

Path Rewriting

HTTPProxy supports rewriting the HTTP request URL path prior to delivering the request to the backend service. Rewriting is performed after a routing decision has been made, and never changes the request destination.

The pathRewritePolicy field specifies how the path prefix should be rewritten. The replacePrefix rewrite policy specifies a replacement string for a HTTP request path prefix match. When this field is present, the path prefix that the request matched is replaced by the text specified in the replacement field. If the HTTP request path is longer than the matched prefix, the remainder of the path is unchanged.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: rewrite-example
  namespace: default
spec:
  virtualhost:
    fqdn: rewrite.bar.com
  routes:
  - services:
    - name: s1
      port: 80
    pathRewritePolicy:
      replacePrefix:
      - replacement: /new/prefix

The replacePrefix field accepts an array of possible replacements. When more than one replacePrefix array element is present, the prefix field can be used to disambiguate which replacement to apply.

If no prefix field is present, the replacement is applied to all prefix matches made against the route. If a prefix field is present, the replacement is applied only to routes that have an exactly matching prefix condition. Specifying more than one replacePrefix entry is mainly useful when a HTTPProxy document is included into multiple parent documents.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: rewrite-example
  namespace: default
spec:
  virtualhost:
    fqdn: rewrite.bar.com
  routes:
  - services:
    - name: s1
      port: 80
    conditions:
    - prefix: /v1/api
    pathRewritePolicy:
      replacePrefix:
      - prefix: /v1/api
        replacement: /app/api/v1
      - prefix: /
        replacement: /app

Header Policy

HTTPProxy supports rewriting HTTP request and response headers. The Set operation sets a HTTP header value, creating it if it doesn’t already exist or overwriting it if it does. The Remove operation removes a HTTP header. The requestHeadersPolicy field is used to rewrite headers on a HTTP request, and the responseHeadersPolicy is used to rewrite headers on a HTTP response. These fields can be specified on a route or on a specific service, depending on the rewrite granularity you need.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: header-rewrite-example
spec:
  virtualhost:
    fqdn: header.bar.com
  routes:
  - services:
    - name: s1
      port: 80
    requestHeadersPolicy:
      set:
      - name: Host
        value: external.dev
      remove:
      - Some-Header
      - Some-Other-Header

ExternalName

HTTPProxy supports routing traffic to ExternalName service types. Contour looks at the spec.externalName field of the service and configures the route to use that DNS name instead of utilizing EDS.

There’s nothing specific in the HTTPProxy object that needs to be configured other than referencing a service of type ExternalName. HTTPProxy supports the requestHeadersPolicy field to rewrite the Host header after first handling a request and before proxying to an upstream service. This field can be used to ensure that the forwarded HTTP request contains the hostname that the external resource is expecting.

NOTE: The ports are required to be specified.

# httpproxy-externalname.yaml
apiVersion: v1
kind: Service
metadata:
  labels:
    run: externaldns
  name: externaldns
  namespace: default
spec:
  externalName: foo-basic.bar.com
  ports:
  - name: http
    port: 80
    protocol: TCP
    targetPort: 80
  type: ExternalName

Proxy to external resource

To proxy to another resource outside the cluster (e.g. A hosted object store bucket for example), configure that external resource in a service type externalName. Then define a requestHeadersPolicy which replaces the Host header with the value of the external name service defined previously. Finally, if the upstream service is served over TLS, set the protocol field on the service to tls or annotate the external name service with: projectcontour.io/upstream-protocol.tls: 443,https assuming your service had a port 443 and name https.

HTTPProxy inclusion

HTTPProxy permits the splitting of a system’s configuration into separate HTTPProxy instances using inclusion.

Inclusion, as the name implies, allows for one HTTPProxy object to be included in another, optionally with some conditions inherited from the parent. Contour reads the inclusion tree and merges the included routes into one big object internally before rendering Envoy config. Importantly, the included HTTPProxy objects do not have to be in the same namespace, so this is functionally the same as the delegation feature of the now-deprecated IngressRoute.

Each tree of HTTPProxy starts with a root, the top level object of the configuration for a particular virtual host. Each root HTTPProxy defines a virtualhost key, which describes properties such as the fully qualified name of the virtual host, TLS configuration, etc.

HTTPProxies included from the root must not contain a virtualhost key. Root objects cannot include other roots either transitively or directly. This permits the owner of an HTTPProxy root to allow the inclusion of a portion of the route space inside a virtual host, and to allow that route space to be further subdivided with inclusions. Because the path is not necessarily used as the only key, the route space can be multi-dimensional.

Conditions and Inclusion

Like Routes, Inclusion may specify a set of conditions. These conditions are added to any conditions on the routes included. This process is recursive.

Conditions are sets of individual condition statements, for example prefix: /blog is the condition that the matching request’s path must start with /blog. When conditions are combined through inclusion Contour merges the conditions inherited via inclusion with any conditions specified on the route. This may result in duplicates, for example two prefix: conditions, or two header match conditions with the same name and value. To resolve this Contour applies the following logic.

  • prefix: conditions are concatenated together in the order they were applied from the root object. For example the conditions, prefix: /api, prefix: /v1 becomes a single prefix: /api/v1 conditions. Note: Multiple prefixes cannot be supplied on a single set of Route conditions.
  • Proxies with repeated identical header: conditions of type “exact match” (the same header keys exactly) are marked as “Invalid” since they create an un-routable configuration.

Configuring inclusion

Inclusion is a top-level part of the HTTPProxy spec element. It requires one field, name, and has two optional fields:

  • namespace. This will assume the included HTTPProxy is in the same namespace if it’s not specified.
  • a conditions block.

Within the same namespace

HTTPProxies can include other HTTPProxy objects in the namespace by specifying the name of the object and its namespace in the top-level includes block. Note that includes is a list, and so it must use the YAML list construct.

In this example, the HTTPProxy include-root has included the configuration for paths matching /service2 from the HTTPPRoxy named service2 in the same namespace as include-root (the default namespace). It’s important to note that service2 HTTPProxy has not defined a virtualhost property as it is NOT a root HTTPProxy.

# httpproxy-inclusion-samenamespace.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: include-root
  namespace: default
spec:
  virtualhost:
    fqdn: root.bar.com
  includes:
  # Includes the /service2 path from service2 in the same namespace
  - name: www
    namespace: default
    conditions:
    - prefix: /service2
  routes:
    - conditions:
      - prefix: /
      services:
        - name: s1
          port: 80
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: service2
  namespace: default
spec:
  routes:
    - conditions:
      - prefix: / # matches /service2
      services:
        - name: s2
          port: 80
    - conditions:
      - prefix: /blog # matches /service2/blog
      services:
        - name: blog
          port: 80

Virtualhost aliases

To present the same set of routes under multiple dns entries, for example www.example.com and example.com, including a service with a prefix condition of / can be used.

# httpproxy-inclusion-multipleroots.yaml
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: multiple-root
  namespace: default
spec:
  virtualhost:
    fqdn: bar.com
  includes:
  - name: main
    namespace: default
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: multiple-root-www
  namespace: default
spec:
  virtualhost:
    fqdn: www.bar.com
  includes:
  - name: main
    namespace: default
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: main
  namespace: default
spec:
  routes:
  - services:
    - name: s2
      port: 80

Across namespaces

Inclusion can also happen across Namespaces by specifying a namespace in the inclusion. This is a particularly powerful paradigm for enabling multi-team Ingress management.

In this example, the root HTTPProxy has included configuration for paths matching /blog to the blog HTTPProxy object in the marketing namespace.

# httpproxy-inclusion-across-namespaces.yaml
---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: namespace-include-root
  namespace: default
spec:
  virtualhost:
    fqdn: ns-root.bar.com
  includes:
  # delegate the subpath, `/blog` to the IngressRoute object in the marketing namespace with the name `blog`
  - name: blog
    namespace: marketing
    conditions:
    - prefix: /blog
  routes:
    - services:
        - name: s1
          port: 80

---
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: blog
  namespace: marketing
spec:
  routes:
    - services:
        - name: s2
          port: 80

Orphaned HTTPProxy children

It is possible for HTTPProxy objects to exist that have not been delegated to by another HTTPProxy. These objects are considered “orphaned” and will be ignored by Contour in determining ingress configuration.

Restricted root namespaces

HTTPProxy inclusion allows for Administrators to limit which users/namespaces may configure routes for a given domain, but it does not restrict where root HTTPProxy may be created. Contour has an enforcing mode which accepts a list of namespaces where root HTTPProxy are valid. Only users permitted to operate in those namespaces can therefore create HTTPProxy with the virtualhost field.

This restricted mode is enabled in Contour by specifying a command line flag, --root-namespaces, which will restrict Contour to only searching the defined namespaces for root HTTPProxy. This CLI flag accepts a comma separated list of namespaces where HTTPProxy are valid (e.g. --root-namespaces=default,kube-system,my-admin-namespace).

HTTPProxy with a defined virtualhost field that are not in one of the allowed root namespaces will be flagged as invalid and will be ignored by Contour.

Additionally, when defined, Contour will only watch for Kubernetes secrets in these namespaces ignoring changes in all other namespaces. Proper RBAC rules should also be created to restrict what namespaces Contour has access matching the namespaces passed to the command line flag. An example of this is included in the examples directory and shows how you might create a namespace called root-httproxies.

NOTE: The restricted root namespace feature is only supported for HTTPProxy CRDs. --root-namespaces does not affect the operation of v1beta1.Ingress objects

TCP Proxying

HTTPProxy supports proxying of TLS encapsulated TCP sessions.

Note: The TCP session must be encrypted with TLS. This is necessary so that Envoy can use SNI to route the incoming request to the correct service.

TLS Termination at the edge

If spec.virtualhost.tls.secretName is present then that secret will be used to decrypt the TCP traffic at the edge.

# httpproxy-tls-termination.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: example
  namespace: default
spec:
  virtualhost:
    fqdn: tcp.example.com
    tls:
      secretName: secret
  tcpproxy:
    services:
    - name: tcpservice
      port: 8080
    - name: otherservice
      port: 9999
      weight: 20

The spec.tcpproxy key indicates that this root HTTPProxy will forward the de-encrypted TCP traffic to the backend service.

TLS passthrough to the backend service

If you wish to handle the TLS handshake at the backend service set spec.virtualhost.tls.passthrough: true indicates that once SNI demuxing is performed, the encrypted connection will be forwarded to the backend service. The backend service is expected to have a key which matches the SNI header received at the edge, and be capable of completing the TLS handshake. This is called SSL/TLS Passthrough.

# httpproxy-tls-passthrough.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: example
  namespace: default
spec:
  virtualhost:
    fqdn: tcp.example.com
    tls:
      passthrough: true
  tcpproxy:
    services:
    - name: tcpservice
      port: 8080
    - name: otherservice
      port: 9999
      weight: 20

TCPProxy delegation

There can be at most one TCPProxy stanza per root HTTPProxy, however that TCPProxy does not need to be defined in the root HTTPProxy object. HTTPProxy authors can delegate the configuration of a TCPProxy to the TCPProxy configuration defined in a HTTPProxy child object.

# httpproxy-parent-termination.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: parent
  namespace: default
spec:
  virtualhost:
    fqdn: tcp.example.com
    tls:
      secretName: secret
  tcpproxy:
    include:
      name: child
      namespace: app
---
# httpproxy-child-termination.yaml
apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: child
  namespace: app
spec:
  tcpproxy:
    services:
    - name: tcpservice
      port: 8080
    - name: otherservice
      port: 9999
      weight: 20

In this example default/parent delegates the configuration of the TCPProxy services to app/child.

TCP Proxy health checking

Active health checking can be configured on a per route basis. Contour supports TCP health checking and can be configured with various settings to tune the behavior.

During TCP health checking Envoy will send a connect-only health check to the upstream Endpoints. It is important to note that these are health checks which Envoy implements and are separate from any other system such as those that exist in Kubernetes.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: tcp-health-check
  namespace: default
spec:
  virtualhost:
    fqdn: health.bar.com
  tcpproxy:
    healthCheckPolicy:
      intervalSeconds: 5
      timeoutSeconds: 2
      unhealthyThresholdCount: 3
      healthyThresholdCount: 5
    services:
      - name: s1-health
        port: 80
      - name: s2-health
        port: 80

TCP Health check policy configuration parameters:

  • intervalSeconds: The interval (seconds) between health checks. Defaults to 5 seconds if not set.
  • timeoutSeconds: The time to wait (seconds) for a health check response. If the timeout is reached the health check attempt will be considered a failure. Defaults to 2 seconds if not set.
  • unhealthyThresholdCount: The number of unhealthy health checks required before a host is marked unhealthy. Note that for http health checking if a host responds with 503 this threshold is ignored and the host is considered unhealthy immediately. Defaults to 3 if not defined.
  • healthyThresholdCount: The number of healthy health checks required before a host is marked healthy. Note that during startup, only a single successful health check is required to mark a host healthy.

Upstream Validation

When defining upstream services on a route, it’s possible to configure the connection from Envoy to the backend endpoint to communicate over TLS. Two configuration items are required, a CA certificate and a SubjectName which are both used to verify the backend endpoint’s identity.

The CA certificate bundle for the backend service should be supplied in a Kubernetes Secret. The referenced Secret must be of type “Opaque” and have a data key named ca.crt. This data value must be a PEM-encoded certificate bundle.

In addition to the CA certificate and the subject name, the Kubernetes service must also be annotated with a Contour specific annotation: projectcontour.io/upstream-protocol.tls: <port> ([see annotations section][9])

Note: This annotation is applied to the Service not the Ingress or HTTPProxy object.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: blog
  namespace: marketing
spec:
  routes:
    - services:
        - name: s2
          port: 80
          validation:
            caSecret: foo-ca-cert
            subjectName: foo.marketing

Client Certificate Validation

It is possible to protect the backend service from unauthorized external clients by requiring the client to present a valid TLS certificate. Envoy will validate the client certificate by verifying that it is not expired and that a chain of trust can be established to the configured trusted root CA certificate. Only those requests with a valid client certificate will be accepted and forwarded to the backend service.

apiVersion: projectcontour.io/v1
kind: HTTPProxy
metadata:
  name: with-client-auth
spec:
  virtualhost:
    fqdn: www.example.com
    tls:
      secretName: secret
      clientValidation:
        caSecret: client-root-ca
  routes:
    - services:
        - name: s1
          port: 80

The preceding example enables validation by setting the optional clientValidation attribute. Its mandatory attribute caSecret contains a name of an existing Kubernetes Secret that must be of type “Opaque” and have a data key named ca.crt. The data value of the key ca.crt must be a PEM-encoded certificate bundle and it must contain all the trusted CA certificates that are to be used for validating the client certificate.

Status Reporting

There are many misconfigurations that could cause an HTTPProxy or delegation to be invalid. To aid users in resolving these issues, Contour updates a status field in all HTTPProxy objects. In the current specification, invalid HTTPProxy are ignored by Contour and will not be used in the ingress routing configuration.

If an HTTPProxy object is valid, it will have a status property that looks like this:

status:
  currentStatus: valid
  description: valid HTTPProxy

If the HTTPProxy is invalid, the currentStatus field will be invalid and the description field will provide a description of the issue.

As an example, if an HTTPProxy object has specified a negative value for weighting, the HTTPProxy status will be:

status:
  currentStatus: invalid
  description: "route '/foo': service 'home': weight must be greater than or equal to zero"

Some examples of invalid configurations that Contour provides statuses for:

  • Negative weight provided in the route definition.
  • Invalid port number provided for service.
  • Prefix in parent does not match route in delegated route.
  • Root HTTPProxy created in a namespace other than the allowed root namespaces.
  • A given Route of an HTTPProxy both delegates to another HTTPProxy and has a list of services.
  • Orphaned route.
  • Delegation chain produces a cycle.
  • Root HTTPProxy does not specify fqdn.
  • Multiple prefixes cannot be specified on the same set of route conditions.
  • Multiple header conditions of type “exact match” with the same header key.

[9]: {% link docs/{{page.version}}/annotations.md %} [10]: /docs/{{page.version}}/api/#projectcontour.io/v1.Service [11]: configuration.md#fallback-certificate

Related Content

Ready to try Contour?

Read our getting started documentation.