HAProxy Kubernetes Ingress Controller Documentation 1.9

HAProxy Kubernetes Ingress Controller Service annotations

These options can be set in a Kubernetes Service object’s metadata.annotations section to change how requests are routed for a particular service.

check

Enables TCP level health checks on pods and attempts a TCP connection periodically.

Values

  • true
  • false

Default

  • true

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/check: "true"

check-http

Enables HTTP level health checks on pods and sends an HTTP request periodically. The check setting must be true.

Values

  • URI to make HTTP requests to, e.g. /health
  • URI with method, e.g. HEAD /health
  • URI, method and HTTP version, e.g. HEAD /health HTTP/1.1

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/check: "true"
    haproxy.org/check-http: "/health"

check-interval

Sets the interval between health checks when check is enabled.

Values

  • Integer with time unit suffix (1m = 1 minute, 10s = 10 seconds)

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/check: "true"
    haproxy.org/check-interval: "1m"

backend-config-snippet

Defines a group of configuration directives to add directly to a HAProxy backend section.

Values

  • One or more valid HAProxy directives

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/backend-config-snippet: |
      http-send-name-header x-dst-server
      stick-table type string len 32 size 100k expire 30m
      stick on req.cook(sessionid)

Enables persistent connections (sticky sessions) between a client and a pod by inserting a cookie into the client’s browser that is used to remember which backend pod they connected to before. Dynamic cookies are used by default via a dynamic-cookie-key in order to support sticky sessions across multiple Ingress Controller instances/replicas.

  • This will insert the following cookie configuration in the corresponding backend cookie <cookie-name> insert indirect nocache dynamic with <cookie-name> the value of this annotation.

Values

  • A name for the cookie

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/cookie-persistence: "mycookie"

forwarded-for

Adds the X-Forwarded-For HTTP header to requests to capture and relay the client’s source IP address to backend pods.

Values

  • true
  • false

Default

  • true

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/forwarded-for: "true"

load-balance

Sets the load-balancing algorithm to use.

Values

  • roundrobin
  • static-rr
  • leastconn
  • first
  • source
  • uri
  • url_param
  • hdr([name])
  • random
  • random([draws])
  • rdp-cookie
  • rdp-cookie([name])

Default

  • roundrobin

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/load-balance: "leastconn"

pod-maxconn

Sets the maximum number of concurrent connections (maxconn) on a backend server (application pod).

  • NB, If multiple HAProxy instances are running, the maxconn will be pod-maxconn number devided by the number of haproxy instances.

Values

  • An integer setting the maximum number of concurrent backend connections

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/pod-maxconn: 30

route-acl

Insert a custom route (use_backend rule) to route ingress traffic to the annotated service based on the provided ACL.

  • In order for the service to be handled by the Ingress Controller, it is still mandatory to put it in an ingress rule. Using only route-acl won’t be enough.

Values

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/route-acl: cookie(staging) -m found

send-proxy-protocol

Uses the PROXY Protocol when connecting to backend servers.

Values

  • proxy - Uses PROXY v1
  • proxy-v1 - Uses PROXY v1
  • proxy-v2 - Uses PROXY v2
  • proxy-v2-ssl Uses PROXY v2 with SSL information extension
  • proxy-v2-ssl-cn Uses PROXY v2 with SSL and Common Name information extension

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/send-proxy-protocol: proxy-v2

server-ca

Sets the certificate authority for backend servers enabling HAProxy to check backend certificates (TLS authentication) when sending encrypted traffic to the kubernetes applications.

  • When used with server-crt.
  • The secret must use ‘tls.crt’ key.

Values

  • Secret path following namespace/secretname format.

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/server-ca: "ns1/ca"

server-crt

Specifies the path of a secret containing a certificate that HAProxy can provide during TLS communication with the backend servers.

  • The secret must use ‘tls.key’ and ‘tls.crt’ keys.
  • When used with server-ca.

Values

  • Secret path following namespace/secretname format.

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/server-crt: "ns1/client"

server-proto

HTTP/1.1 is the default protocol for backend servers communication. Currently, the server-proto annotation supports only “h2” as a value (supporting fcgi is also planned) which transmits HTTP/2 messages in the clear to the backend servers. However, when SSL is enabled on the backend, server-proto is ignored and both HTTP/1.1 and HTTP/2 are advertised via ALPN and transmitted as encrypted messages.

Values

  • h2

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/server-proto: "h2"

server-ssl

Enables SSL to pods.

  • Enable HTTP/2 support for backend severs.

Values

  • true
  • false

Default

  • false

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/server-ssl: "true"

scale-server-slots

Sets the number of server slots to provision in order for HAProxy to scale dynamically with no reload. If this number is greater than the available endpoints/addresses, the remaining slots will be disabled (put on stand-by) and ready to be used. If this number is lower, the remaining endpoints/addresses will be added after scaling the HAProxy backend with a reload.

  • Equivalent old annotations are servers-increment and server-slots

Values

  • Integer value indicating the number of backend servers to provision. Defaults to 42.

Default

  • 42

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/scale-server-slots: "75"

ssl-passthrough

Passes SSL/TLS traffic through at Layer 4 directly to the backend service without Layer 7 inspection.

  • Traffic is proxied in TCP mode which makes unavailable a number of the controller annotations (requiring HTTP mode).

Values

  • true
  • false

Default

  • false

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/ssl-passthrough: "true"

timeout-check

Sets an additional check timeout, but only after a connection has been already established.

Values

  • An integer with a unit of time (1 second = 1s, 1 minute = 1m, 1h = 1 hour)

Default

  • no default value

Example

apiVersion: v1
kind: Service
metadata:
  labels:
    run: web
  name: web
  annotations:

    haproxy.org/timeout-check: 5s

Next up

Target multiple ingress controllers via IngressClassName