Configuration keys

Configuration keys are entry point configurations that allow users and admins to dynamically fine-tune HAProxy status. HAProxy Ingress reads configuration keys from Kubernetes resources, and this can be done in a couple of ways:

• Globally, from a ConfigMap
• Per IngressClass, from a ConfigMap linked in the IngressClass’ parameters field
• Per Ingress, configuring or annotating Ingress resources
• Per backend, annotating Service resources

The list above also describes the precedence if the same configuration key is used in more than one resource: Global configurations can be overridden by IngressClass configurations, that can be overriden by Ingress resource configurations and so on. This hierarchy creates a flexible model, where commonly used configurations can be made in a higher level and overriden by local changes.

The following sections describe in a few more details how HAProxy Ingress classifies an Ingress to be part of the final configuration, and how it reads the configuration from Kubernetes resources.

Class matter

HAProxy Ingress by default does not listen to Ingress resources, until one or more of the following conditions are met:

• Ingress resources have the annotation kubernetes.io/ingress.class with the value haproxy
• Ingress resources have its ingressClassName field assigning an IngressClass resource whose controller name is haproxy-ingress.github.io/controller
• HAProxy Ingress was started with --watch-ingress-without-class command-line option

See Ingress Class command-line doc for customization options.

The first two options give more control on which Ingress resources should be part of the final configuration. Class annotation and the IngressClass name can be changed on a running controller, the configuration will be adjusted on the fly to reflect the new status. If both options are configured in an Ingress resource, and they conflict - i.e. one of them says the controller belongs to HAProxy Ingress and the other says that it does not belong - the annotation value wins and a warning is logged.

Adding a class annotation or defining an IngressClass name means “classify” an Ingress resource. The third and latest option asks HAProxy Ingress to also add “unclassified” Ingress to the final configuration - i.e. add Ingress resources that does not have the kubernetes.io/ingress.class annotation and also does not have the ingressClassName field. Note that this is a new behavior since v0.12. Up to v0.11 HAProxy Ingress listen to “unclassified” Ingress by default.

Strategies

HAProxy Ingress reads configuration on three distinct ways:

• ConfigMap key/value data. ConfigMaps are assigned either via --configmap command-line option (used by Global options), or via parameters field of an IngressClass
• Annotations from classified Ingress resources and also from Services that these Ingress are linking to
• Spec configurations from classified Ingress resources

HAProxy Ingress follows Ingress v1 spec, so any Ingress spec configuration should work as stated by the Kubernetes documentation.

Annotations and ConfigMap customizations extend the Ingress spec via the configuration keys, and this is what the rest of this documentation page is all about.

The following sections describe in a few more details about configuration strategies.

ConfigMap

ConfigMap key/value options are read in the following conditions:

• Global config, using --configmap command-line option. The installation process configures a Global config ConfigMap named haproxy-ingress in the controller namespace. This is the only way to configure keys from the Global scope. See about scopes later in this page.
• IngressClass config, using its parameters field linked to a ConfigMap declared in the same namespace of the controller. See about IngressClass later in this same section.

A configuration key is used verbatim as the ConfigMap key name, without any prefix. The ConfigMap spec expects a string as the key value, so declare numbers and booleans as strings, HAProxy Ingress will convert them when needed.

apiVersion: v1
data:
  balance-algorithm: leastconn
  max-connections: "10000"
  ssl-redirect: "true"
kind: ConfigMap
metadata:
  name: haproxy-ingress
  namespace: ingress-controller

Annotation

Annotations are read in the following conditions:

• From classified Ingress resources, see about classification in the Class matter section. Ingresses accept keys from the Host, Backend and Path scopes. See about scopes later in this page.
• From Services that classified Ingress resources are linking to. Services only accept keys from the Backend scope.

A configuration key needs a prefix in front of its name to use as an annotation key. The default prefix is ingress.kubernetes.io, change with the --annotation-prefix command-line option. The annotation value spec expects a string as the key value, so declare numbers and booleans as strings, HAProxy Ingress will convert them when needed.

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  annotations:
    ingress.kubernetes.io/balance-algorithm: roundrobin
    ingress.kubernetes.io/maxconn-server: "500"
    ingress.kubernetes.io/ssl-redirect: "false"
  name: app
  namespace: default
spec:
  ...

IngressClass

IngressClass configurations are read when the ingressClassName field of an Ingress resource links to an IngressClass that configures its parameters field.

The IngressClass’ parameters field currently only accepts ConfigMap resources, and the ConfigMap must be declared in the same namespace of the controller.

The following resources create the same final configuration of the Annotation section above, with the benefit of allowing the reuse of the IngressClass+ConfigMap configuration.

apiVersion: networking.k8s.io/v1beta1
kind: IngressClass
metadata:
  name: my-class
spec:
  controller: haproxy-ingress.github.io/controller
  parameters:
    kind: ConfigMap
    name: my-options

apiVersion: v1
data:
  balance-algorithm: roundrobin
  maxconn-server: "500"
  ssl-redirect: "false"
kind: ConfigMap
metadata:
  name: my-options
  namespace: ingress-controller

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  name: app
  namespace: default
spec:
  ingressClassName: my-class
  ...

Updates

Changes to any configuration in any classified Ingress resources (annotations or spec), Service resources (annotations) or any referenced ConfigMap will reflect in the update of the final HAProxy configuration.

If the new state cannot be dynamically applied and requires HAProxy to be reloaded, this will happen preserving the in progress requests and the long running connections.

Fragmentation

Ingress resources can be fragmented in order to add distinct configurations to distinct routes. For example:

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  name: app-front
spec:
  rules:
  - host: app.local
    http:
      paths:
      - path: /
        backend:
          serviceName: frontend
          servicePort: 8080

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  annotations:
    ingress.kubernetes.io/rewrite-target: /
  name: app-back
spec:
  rules:
  - host: app.local
    http:
      paths:
      - path: /api
        backend:
          serviceName: backend
          servicePort: 8080

HAProxy Ingress will merge all the resources, so there is no difference if the configuration is in the same or in distinct Ingress. Distinct Ingress however might lead to conflicting configuration, more about conflict in the scope section below.

There is no hard limit to the number of Ingresses or Services - clusters with tens of thousands of Ingress and Service resources report to work smoothly and fast with HAProxy Ingress.

Scope

HAProxy Ingress configuration keys may be in one of four distinct scopes: Global, Host, Backend, Path. A scope defines where a configuration key can be declared and how it interacts with Ingress and Service resources.

Configuration keys declared in Ingress resources might conflict. More about the scenarios in the Host and Backend scopes below. A warning will be logged in the case of a conflict, and the used value will be of the Ingress resource that was created first.

Global

Defines configuration keys that apply for all hostnames and backend services, and should be declared only in the Global config ConfigMap resource. Configuration keys of the Global scope declared as Ingress or Service annotations, and also in the IngressClass ConfigMap are ignored. Configuration keys of the Global scope never conflict.

Host

Defines configuration keys that bind to the hostname. Configuration keys of the host scope can be declared in any ConfigMap, or in any Ingress resource. A conflict happens when the same host configuration key with distinct values are declared in distinct Ingress resources but to the same hostname.

Backend

Defines configuration keys that bind to the Service resource, which is converted to a HAProxy backend after the configuration parsing. Configuration keys of the backend scope can be declared in any ConfigMap or as Ingress or Service annotation. A conflict happens when the same backend configuration key with distinct values are declared in distinct Ingress resources but to the same Service or HAProxy backend.

Path

Defines configuration keys that bind to the hostname and the HTTP path. Configuration keys of the Path scope can be declared in any ConfigMap as a default value, or as Ingress or Service annotation. Configuration keys of the Path scope never conflict.

Keys

The table below describes all supported configuration keys.

Acme

Configures dynamic options used to authorize and sign certificates against a server which implements the acme protocol, version 2.

The popular Let’s Encrypt certificate authority implements acme-v2.

Supported acme configuration keys:

• acme-emails: mandatory, a comma-separated list of emails used to configure the client account. The account will be updated if this option is changed.
• acme-endpoint: mandatory, endpoint of the acme environment. v2-staging and v02-staging are alias to https://acme-staging-v02.api.letsencrypt.org, while v2 and v02 are alias to https://acme-v02.api.letsencrypt.org.
• acme-expiring: how many days before expiring a certificate should be considered old and should be updated. Defaults to 30 days.
• acme-shared: defines if another certificate signer is running in the cluster. If false, the default value, any request to /.well-known/acme-challenge/ is sent to the local acme server despite any ingress object configuration. Otherwise, if true, a configured ingress object would take precedence.
• acme-terms-agreed: mandatory, it should be defined as true, otherwise certificates won’t be issued.
• cert-signer: defines the certificate signer that should be used to authorize and sign new certificates. The only supported value is "acme". Add this config as an annotation in the ingress object that should have its certificate managed by haproxy-ingress and signed by the configured acme environment. The annotation kubernetes.io/tls-acme: "true" is also supported if the command-line option --acme-track-tls-annotation is used.

Minimum setup

The command-line option --acme-server need to be declared to start the local server and the work queue used to authorize and sign new certificates. See other command-line options here.

The following configuration keys are mandatory: acme-emails, acme-endpoint, acme-terms-agreed.

A cluster-wide permission to create and update the secrets resources should also be made.

How it works

All haproxy-ingress instances should declare --acme-server command-line option, which will start a local server to answer acme challenges, a work queue to enqueue the domain authorization and certificate signing, and will also start a leader election to define which haproxy-ingress instance should perform authorizations and certificate signing.

The haproxy-ingress leader tracks ingress objects that declares the annotation ingress.kubernetes.io/cert-signer with value acme and a configured secret name for TLS certificate. The annotation kubernetes.io/tls-acme with value "true" will also be used if the command-line option --acme-track-tls-annotation is declared. The secret does not need to exist. A new certificate will be issued if the certificate is old, the secret does not exist or has an invalid certificate, or the domains of the certificate doesn’t cover all the domains configured in the ingress.

Every 24h or the duration configured in the --acme-check-period, and also when the leader changes, all the certificates from all the tracked ingress will be verified. The certificate is also verified whenever the list of the domains or the secret name changes, so the periodic check will, in fact, only issue new certificates when there is 30 days or less to the certificate expires. This duration can be changed with acme-expiring configuration key.

If an authorization fails, the certificate request is re-enqueued to be tried again after 5m. This duration can be changed with --acme-fail-initial-duration command-line option. If the request fails again, it will be re-enqueued after the double of the time, in this case, after 10m. The duration will exponentially increase up to 8h or the duration defined by the command-line option --acme-fail-max-duration. The request will continue in the work queue until it is successfully processed and stored, or when the ingress object is untracked, either removing the annotation, removing the secret name or removing the ingress object itself.

See also:

• acme command-line options doc.

Affinity

Configure if HAProxy should maintain client requests to the same backend server.

• affinity: the only supported option is cookie. If declared, clients will receive a cookie with a hash of the server it should be fidelized to.
• cookie-key: defines a secret key used with the IP address and port number of a backend server to dynamically create a cookie to that server. Defaults to Ingress if not provided.
• session-cookie-dynamic: indicates whether or not dynamic cookie value will be used. With the default of true, a cookie value will be generated by HAProxy using a hash of the server IP address, TCP port, and dynamic cookie secret key. When false, the server name will be used as the cookie name. Note that setting this to false will have no impact if use-resolver is set.
• session-cookie-keywords: additional options to the cookie option like nocache, httponly. For the sake of backwards compatibility the default is indirect nocache httponly if not declared and strategy is insert.
• session-cookie-name: the name of the cookie. INGRESSCOOKIE is the default value if not declared.
• session-cookie-preserve: indicates whether the session cookie will be set to preserve mode. If this mode is enabled, haproxy will allow backend servers to use a Set-Cookie HTTP header to emit their own persistence cookie value, meaning the backend servers have knowledge of which cookie value should route to which server. Since the cookie value is tightly coupled with a particular backend server in this scenario, this mode will cause dynamic updating to understand that it must keep the same cookie value associated with the same backend server. If this is disabled, dynamic updating is free to assign servers in a way that can make their cookie value no longer matching.
• session-cookie-same-site: if true, adds the SameSite=None; Secure attributes, which configures the browser to send the persistence cookie with both cross-site and same-site requests. The default value is false, which means only same-site requests will send the persistence cookie.
• session-cookie-shared: defines if the persistence cookie should be shared between all domains that uses this backend. Defaults to false. If true the Set-Cookie response will declare all the domains that shares this backend, indicating to the HTTP agent that all of them should use the same backend server.
• session-cookie-strategy: the cookie strategy to use (insert, rewrite, prefix). insert is the default value if not declared.
• session-cookie-value-strategy: the strategy to use to calculate the cookie value of a server (server-name, pod-uid). server-name is the default if not declared, and indicates that the cookie will be set based on the name defined in backend-server-naming. pod-uid indicates that the cookie will be set to the UID of the pod running the target server.

Note for dynamic-scaling users only, v0.5 or older: the hash of the server is built based on it’s name. When the slots are scaled down, the remaining servers might change it’s server name on HAProxy configuration. In order to circumvent this, always configure the slot increment at least as much as the number of replicas of the deployment that need to use affinity. This limitation was removed on v0.6.

See also:

• https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Set-Cookie
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-cookie
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-cookie
• https://www.haproxy.com/blog/load-balancing-affinity-persistence-sticky-sessions-what-you-need-to-know/
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#dynamic-cookie-key

Agent check

Allows HAProxy agent checks to be defined for a backend. This is an auxiliary check that is run independently of a regular health check and can be used to control the reported status of a server as well as the weight to be used for load balancing.

• agent-check-port: Defines the port on which the agent is listening. This option is required in order to use an agent check.
• agent-check-addr: Defines the address for agent checks. If omitted, the server address will be used.
• agent-check-interval: Defines the interval between agent checks. If omitted, the default of 2 seconds will be used.
• agent-check-send: Defines a string to be sent to the agent upon connection.

The following limitations are known when using agent-check to change the weight of a backend server:

• If using drain-support, the backend server will have its initial weight defined as 0 (zero) if the server is terminating when haproxy is restarted, making the weight update useless
• Blue/green annotation might be dynamically applied, which will temporarily overwrite the weight defined from the agent

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-agent-check
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-agent-port
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-agent-inter
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-agent-send

Allowlist

Defines a comma-separated list of source IPs or CIDRs allowed or denied to connect. The default behavior is to allow all source IPs if neither the allow list nor the deny list are declared. The lists support IPv4 and IPv6.

This is a path scoped configuration: distinct paths in the same hostname can have distinct configurations. However this doesn’t happen if the backend has ssl-passthrough, which uses HAProxy’s TCP mode, in this case the allow and deny lists act as a backend scoped config.

Since v0.12 IPs or CIDRs can be prefixed with !, which means an exception to the rule, so an allow list with "10.0.0.0/8,!10.100.0.0/16" will allow only IPs from the range 10.x.x.x, except the range 10.100.x.x which will continue to be denied.

• allowlist-source-range: Used to deny requests by default, allowing only the IPs and CIDRs in the list, except IPs and CIDRs prefixed with ! which will continue to be denied. whitelist-source-range is an alias to preserve backward compatibility, and will be ignored if allowlist-source-range is declared.
• denylist-source-range: Used to allow requests by default, denying only the IPs and CIDRs in the list, except IPs and CIDRs prefixed with ! which will continue to be allowed.

Allowlist and denylist can be used together. The request will be denied if the configurations overlap and a source IP matches both the allowlist and denylist.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4.2-http-request%20deny

App root

Defines a distinct application root path. HAProxy will redirect requests to the configured path, using 302 status code, when the HTTP client sends a request to the root context of the configured domain. app-root key binds to the root context path, so it needs to be declared in the same Ingress that configures it.

Auth

Configures authentication options.

• auth-type: Defines the authentication type. Currently the only supported option is basic.
• auth-realm: Optional, configures the authentication realm string. localhost will be used if not provided.
• auth-secret: A secret name with users and passwords used to configure basic authentication. The secret can be in the same namespace of the Ingress resource, or any other namespace if cross namespace is enabled. Secret in the same namespace does not need to be prepended with namespace/.

The secret referenced by auth-secret should have a key named auth with users and passwords, one per line. The following two formats are supported and both are supported in the same secret:

• <user>::<password>: User and password are separated by 2 (two) colons. The password will be copied verbatim, stored in the configuration file in an insecure way.
• <user>:<password-hash>: User and password are separated by 1 (one) colon. This syntax needs a password hash that can be generated with mkpasswd.

See also:

• –allow-cross-namespace command-line option
• Auth TLS configuration keys
• OAuth configuration keys
• Auth Basic example page

Auth TLS

Configure client authentication with X509 certificate. The following headers are added to the request:

• X-SSL-Client-SHA1: Hex encoding of the SHA-1 fingerprint of the X509 certificate. The default output uses uppercase hexadecimal digits, configure ssl-fingerprint-lower to true to use lowercase digits instead.
• X-SSL-Client-DN: Distinguished name of the certificate
• X-SSL-Client-CN: Common name of the certificate

The prefix of the header names can be configured with ssl-headers-prefix key. The default value is to X-SSL, which will create a X-SSL-Client-DN header with the DN of the certificate.

The following keys are supported:

• auth-tls-cert-header: If true HAProxy will add X-SSL-Client-Cert http header with a base64 encoding of the X509 certificate provided by the client. Default is to not provide the client certificate.
• auth-tls-error-page: Optional URL of the page to redirect the user if he doesn’t provide a certificate or the certificate is invalid.
• auth-tls-secret: Mandatory secret name with ca.crt key providing all certificate authority bundles used to validate client certificates. Since v0.9, an optional ca.crl key can also provide a CRL in PEM format for the server to verify against.
• auth-tls-strict: Defines if a wrong or incomplete configuration, eg missing secret with ca.crt, should forbid connection attempts. If false, the default value, a wrong or incomplete configuration will ignore the authentication config, allowing anonymous connection. If true, a strict configuration is used: all requests will be rejected with HTTP 495 or 496, or redirected to the error page if configured, until a proper ca.crt is provided. Strict configuration will only be used if auth-tls-secret has a secret name and auth-tls-verify-client is missing or is not configured as off.
• auth-tls-verify-client: Optional configuration of Client Verification behavior. Supported values are off, on, optional and optional_no_ca. The default value is on if a valid secret is provided, off otherwise.
• ssl-fingerprint-lower: Defines if the certificate fingerprint should be in lowercase hexadecimal digits. The default value is false, which uses uppercase digits.
• ssl-headers-prefix: Configures which prefix should be used on HTTP headers. Since RFC 6648 X- prefix on unstandardized headers changed from a convention to deprecation. This configuration allows to select which pattern should be used on header names.

See also:

• example page.

Backend protocol

Defines the HTTP protocol version of the backend. Note that HTTP/2 is only supported if HTX is enabled. A case insensitive match is used, so either h1 or H1 configures HTTP/1 protocol. A non SSL/TLS configuration does not overrides secure-backends, so h1 and secure-backends true will still configures SSL/TLS.

Options:

• h1: the default value, configures HTTP/1 protocol. http is an alias to h1.
• h1-ssl: configures HTTP/1 over SSL/TLS. https is an alias to h1-ssl.
• h2: configures HTTP/2 protocol. grpc is an alias to h2.
• h2-ssl: configures HTTP/2 over SSL/TLS. grpcs is an alias to h2-ssl.

See also:

• use-htx configuration key to enable HTTP/2 backends.
• secure-backend configuration keys to configure optional client certificate and certificate authority bundle of SSL/TLS connections.
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-proto

Backend server naming

Configures how to name backend servers.

• sequence: Names backend servers with a prefixed number sequence: srv001, srv002, and so on. This is the default configuration and the preferred option if dynamic udpate is used. seq is an alias to sequence.
• pod: Uses the k8s pod name as the backend server name. This option doesn’t work on backends whose service-upstream is true, falling back to sequence.
• ip: Uses target’s <ip>:<port> as the server name.

Balance algorithm

Defines a valid HAProxy load balancing algorithm. The default value is roundrobin.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-balance

Bind

Configures listening IP and port for HTTP/s incoming requests. These configuration keys have backward compatibility with Bind IP addr, Bind port and Fronting proxy keys. The bind configuration keys in this section have precedente if declared.

Any HAProxy supported option can be used, this will be copied verbatim to the bind keyword. See HAProxy bind keyword doc.

Configuration examples:

• bind-http: ":::80" and bind-https: ":::443": Listen all IPv6 addresses
• bind-http: ":80,:::80" and bind-https: ":443,:::443": Listen all IPv4 and IPv6 addresses
• bind-https: ":443,:8443": accept https connections on 443 and also 8443 port numbers

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-bind
• Bind IP addr
• Bind port

Bind IP addr

Define listening IPv4/IPv6 address on public HAProxy frontends. Since v0.10 the default value changed from * to an empty string, which haproxy interprets in the same way and binds on all IPv4 address.

• bind-ip-addr-tcp: IP address of all TCP services declared on tcp-services command-line option.
• bind-ip-addr-healthz: IP address of the health check URL.
• bind-ip-addr-http: IP address of all HTTP/s frontends, port :80 and :443, and also https-to-http-port if declared.
• bind-ip-addr-prometheus: IP address of the haproxy’s internal Prometheus exporter.
• bind-ip-addr-stats: IP address of the statistics page. See also stats-port.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-bind
• Bind
• Bind port

Bind port

• healthz-port: Define the port number HAProxy should listen to in order to answer for health checking requests. Use /healthz as the request path.
• http-port: Define the port number of unencripted HTTP connections.
• https-port: Define the port number of encripted HTTPS connections.
• prometheus-port: Define the port number of the haproxy’s internal Prometheus exporter. Defaults to not create the listener. A listener without being scraped does not use system resources, except for the listening port. The internal exporter supports scope filter as a query string, eg /metrics?scope=frontend&scope=backend will only export frontends and backends. See the full description in the HAProxy’s Prometheus exporter doc.

See also:

• Bind configuration key
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-monitor-uri (healthz-port)
• https://git.haproxy.org/?p=haproxy-2.0.git;a=blob;f=contrib/prometheus-exporter/README;hb=HEAD (prometheus-port)

Blue-green

Configure backend server groups based on the weight of the group - blue/green balance - or a group selection based on http header or cookie value - blue/green selector.

Both blue/green configurations can be used together: if the http header or cookie isn’t provided or doesn’t match a group, the blue/green balance will be used.

Blue/green reads endpoint weight from the pod lister. However the --disable-pod-list command-line option can be safely used to save some memory on clusters with a huge amount of pods. If pod list is disabled, pods are read straight from the k8s api, only when needed, without changing blue/green behavior.

See below the description of the two blue/green configuration options.

Blue/green balance

Configures weight of a blue/green deployment. The annotation accepts a comma separated list of label name/value pair and a numeric weight. Concatenate label name, label value and weight with an equal sign, without spaces. The label name/value pair will be used to match corresponding pods or deploys. There is no limit to the number of label/weight balance configurations.

The endpoints of a single backend are selected using service selectors, which also uses labels. Because of that, in order to use blue/green deployment, the deployment, daemon set or replication controller template should have at least two label name/value pairs - one that matches the service selector and another that matches the blue/green selector.

• blue-green-balance: comma separated list of labels and weights
• blue-green-deploy: deprecated on v0.7, this is an alias to blue-green-balance.
• blue-green-mode: defaults to deploy on v0.7, defines how to apply the weights, might be pod or deploy

The following configuration group=blue=1,group=green=4 will redirect 20% of the load to the group=blue group and 80% of the load to group=green group.

Applying the weights depends on the blue/green mode. v0.6 has only pod mode which means that every single pod receives the same weight as configured on blue/green balance. This means that a balance configuration with 50% to each group will redirect twice as much requests to a backend that has the double of replicas. v0.7 has also deploy mode which rebalance the weights based on the number of replicas of each deployment.

In short, regarding blue/green mode: use pod if you want to redirect more requests to a deployment updating the number of replicas; use deploy if you want to control the load of each side updating the blue/green balance annotation.

Value of 0 (zero) can also be used as weight. This will let the endpoint configured in the backend accepting persistent connections - see affinity - but will not participate in the load balancing. The maximum weight value is 256.

Blue/green selector

Configures header or cookie name and also a pod label name used to tag the group of backend servers.

• blue-green-cookie: the CookieName:LabelName pair
• blue-green-header: the HeaderName:LabelName pair

The CookieName or HeaderName is the name of the http cookie or header used in the request to match a group name. The LabelName is the name of the pod label used to read the group name of the backend server.

The following configuration X-Server:group on blue-green-header configures HAProxy to try to match a backend server based on the value of its label group. A request with header X-Server: green will match a pod labeled group=green. Cookie configuration follows the same rules.

The name of the header and the label follow the k8s label naming convention: must consist of alphanumeric characters, -, _ or ., and must start and end with an alphanumeric character.

Both cookie and header based configurations can be used together in the same backend (k8s service), provided that the label name is the same. If the request uses the configured header and cookie, the header will take precedence, and the cookie would be used if the header value provided doesn’t match a healthy backend server.

Note that blue/green selector should be used only on controlled testing scenarios because it doesn’t provide a proper load balancing: the first healthy backend server that match header or cookie configuration will be used despite if a proper load balance algorithm would choose another one. This can be changed in the future. Blue/green balance doesn’t have this limitation and properly uses the chosen load balance algorithm.

See also:

• example page.
• disable-pod-list command-line option doc.
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-weight (weight based balance)
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-use-server (use-server based selector)

Configuration snippet

Add HAProxy configuration snippet to the configuration file. Use multiline content to add more than one line of configuration.

Examples - ConfigMap:

    config-global: |
      tune.bufsize 32768

    config-defaults: |
      option redispatch

    config-frontend: |
      capture request header X-User-Id len 32

Annotation:

    annotations:
      ingress.kubernetes.io/config-backend: |
        acl bar-url path /bar
        http-request deny if bar-url

The following keys add a configuration snippet to the …:

• config-backend: … HAProxy backend section.
• config-global: … end of the HAProxy global section.
• config-defaults: … end of the HAProxy defaults section.
• config-frontend: … HAProxy frontend sections.

Connection

Configuration of connection limits.

• max-connections: Define the maximum concurrent connections on all proxies. Defaults to 2000 connections, which is also the HAProxy default configuration.
• maxconn-server: Defines the maximum concurrent connections each server of a backend should receive. If not specified or a value lesser than or equal zero is used, an unlimited number of connections will be allowed. When the limit is reached, new connections will wait on a queue.
• maxqueue-server: Defines the maximum number of connections should wait in the queue of a server. When this number is reached, new requests will be redispached to another server, breaking sticky session if configured. The queue will be unlimited if the annotation is not specified or a value lesser than or equal to zero is used.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.2-maxconn (max-connections)
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-maxconn (maxconn-server)
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-maxqueue (maxqueue-server)

CORS

Add CORS headers on OPTIONS http command (preflight) and reponses.

• cors-enable: Enable CORS if defined as true.
• cors-allow-origin: Optional, configures Access-Control-Allow-Origin header which defines the URL that may access the resource. Defaults to *.
• cors-allow-methods: Optional, configures Access-Control-Allow-Methods header which defines the allowed methods. Default value is GET, PUT, POST, DELETE, PATCH, OPTIONS.
• cors-allow-headers: Optional, configures Access-Control-Allow-Headers header which defines the allowed headers. Default value is DNT,X-CustomHeader,Keep-Alive,User-Agent,X-Requested-With,If-Modified-Since,Cache-Control,Content-Type,Authorization.
• cors-allow-credentials: Optional, configures Access-Control-Allow-Credentials header which defines whether or not credentials (cookies, authorization headers or client certificates) should be exposed. Defaults to true.
• cors-max-age: Optional, configures Access-Control-Max-Age header which defines the time in seconds the result should be cached. Defaults to 86400 (1 day).
• cors-expose-headers: Optional, configures Access-Control-Expose-Headers header which defines what headers are allowed to be passed through to the CORS application. Defaults to not add the header.

See also:

• https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS

CPU map

Define how processes/threads map to CPUs. The default value is generated based on nbthread and nbproc.

• cpu-map: Custom override specifying the cpu mapping behaviour in the format described here.
• use-cpu-map: Set to false to prevent any cpu mapping

See also:

• nbthread configuration key
• nbproc configuration key
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-cpu-map

Default Redirect

Define a redirect location of the HAProxy for unknown resources.

• default-backend-redirect: Defines a location in which Ingress should redirect an user if the incoming request doesn’t match any hostname, or the requested path doesn’t match any location within the desired hostname. An internal 404 error page is used if not declared and also if default-backend-service was not configured on command line.

• default-backend-redirect-code: Defines the return code to be used when redirecting a user. Defaults to 302 (Moved Temporarily)

DNS resolvers

Configure dynamic backend server update using DNS service discovery.

The following keys are supported:

• dns-resolvers: Multiline list of DNS resolvers in resolvername=ip:port format
• dns-accepted-payload-size: Maximum payload size announced to the name servers
• dns-timeout-retry: Time between two consecutive queries when no valid response was received, defaults to 1s
• dns-hold-valid: Time a resolution is considered valid. Keep in sync with DNS cache timeout. Defaults to 1s
• dns-hold-obsolete: Time to keep valid a missing IP from a new DNS query, defaults to 0s
• dns-cluster-domain: K8s cluster domain, defaults to cluster.local
• use-resolver: Name of the resolver that the backend should use

See also:

• example page.
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.3.2
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-resolvers
• https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/
• https://kubernetes.io/docs/concepts/services-networking/service/#headless-services

Drain support

Set drain-support to true if you wish to use HAProxy’s drain support for pods that are NotReady (e.g., failing a k8s readiness check) or are in the process of terminating. This option only makes sense with cookie affinity configured as it allows persistent traffic to be directed to pods that are in a not ready or terminating state.

By default, sessions will be redispatched on a failed upstream connection once the target pod is terminated. You can control this behavior by setting drain-support-redispatch flag to false to instead return a 503 failure.

See also:

• disable-pod-list command-line option doc.

Dynamic scaling

The dynamic-scaling option defines if backend updates should always be made starting a new HAProxy instance that will read the new config file (false), or updating the running HAProxy via a Unix socket (true) whenever possible. Despite the configuration, the config files will stay in sync with in memory config. The default value was false up to v0.7 if not declared, changed to true since v0.8.

dynamic-scaling is ignored if the backend uses DNS resolver.

If true HAProxy Ingress will create at least backend-server-slots-increment servers on each backend and update them via a Unix socket without reloading HAProxy. Unused servers will stay in a disabled state. If the change cannot be made via socket, a new HAProxy instance will be started.

Starting on v0.8, a new ConfigMap option slots-min-free can be used to configure the minimum number of free/empty servers per backend. If HAProxy need to be restarted and an backend has less than slots-min-free available servers, another backend-server-slots-increment new empty servers would be created.

Starting on v0.6, dynamic-scaling config will only force a reloading of HAProxy if the number of servers on a backend need to be increased. Before v0.6 a reload will also happen when the number of servers could be reduced.

The following keys are supported:

• dynamic-scaling: Define if dynamic scaling should be used whenever possible
• backend-server-slots-increment: Configures the minimum number of servers, the size of the increment when growing and the size of the decrement when shrinking of each HAProxy backend
• slots-min-free: Configures the minimum number of empty servers a backend should have on every HAProxy restarts

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/management.html#9.3

External

Defines features that can be found in the external haproxy deployment, if an external deployment is used. These options have no effect if using the embedded haproxy.

• external-has-lua: Define as true if the external haproxy has Lua libraries installed in the operating system. Currently only OAuth needs Lua socket installed and will not work if external-has-lua is not enabled.

See also:

• OAuth configuration keys.
• master-socket command-line option

Forwardfor

Define how the X-Forwarded-For header should be handled by haproxy.

Options:

• add: haproxy should generate a X-Forwarded-For header with the source IP address. This is the default option and should be used on untrusted networks. If the request has a XFF header, its value is copied to X-Original-Forwarded-For.
• update: Only on v0.9 and above. haproxy should preserve any X-Forwarded-For header, if provided, updating with the source IP address, which should be a fronting TCP or HTTP proxy/load balancer.
• ignore: do nothing - only send the X-Forwarded-For header if the client provided one, without updating its content.
• ifmissing: add X-Forwarded-For header only if the incoming request doesn’t provide one.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-option%20forwardfor
• https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/X-Forwarded-For

Fronting proxy port

A port number to listen to http requests from a fronting proxy that does the ssl offload, eg haproxy ingress behind a cloud load balancers that manages the TLS certificates. https-to-http-port is an alias to fronting-proxy-port.

fronting-proxy-port and http-port can share the same port number, see below what changes in the behaviour.

use-forwarded-proto defines if haproxy should use X-Forwarded-Proto header to decide how to handle requests made to fronting-proxy-port port number.

If use-forwarded-proto is false, the request takes the https route and is handled as if X-Forwarded-Proto header is https, see below. The actual header content is ignored by haproxy and forwarded to the backend if provided.

If use-forwarded-proto is true, the default value, requests made to fronting-proxy-port port number evaluate the X-Forwarded-Proto header to decide how to handle the request:

• If X-Forwarded-Proto header is https:
  • HAProxy will handle the request just like the ssl-offload was made by HAProxy itself - HSTS header is provided if configured and X-SSL-* headers won’t be changed or removed if provided.
• If X-Forwarded-Proto header is http or any other value except https:
  • HAProxy will redirect scheme to https
• If X-Forwarded-Proto header is missing:
  • If fronting-proxy-port has its own port — HAProxy will redirect scheme to https
  • If fronting-proxy-port shares the HTTP port — the request will be handled as plain http, being redirected to https only if ssl-redirect is true, just like if fronting-proxy-port wasn’t configured.

See also:

• Bind
• Bind port

Headers

Configures a list of HTTP header names and the value it should be configured with. More than one header can be configured using a multi-line configuration value. The name of the header and its value should be separated with a colon and/or any amount of spaces.

The following variables can be used in the value:

• %[namespace]: namespace of the ingress or service
• %[service]: name of the service which received the request

Configuration example:

    annotations:
      ingress.kubernetes.io/headers: |
        x-path: /
        host: %[service].%[namespace].svc.cluster.local

Health check

Controls server health checks on a per-backend basis.

• health-check-uri: If specified, this changes the default TCP health into an HTTP health check.
• health-check-addr: Defines the address for health checks. If omitted, the server addr will be used.
• health-check-port: Defines the port for health checks. If omitted, the server port will be used.
• health-check-interval: Defines the interval between health checks. The default value 2s is used if omitted.
• health-check-rise-count: The number of successful health checks that must occur before a server is marked operational. If omitted, the default value is 2.
• health-check-fall-count: The number of failed health checks that must occur before a server is marked as dead. If omitted, the default value is 3.
• backend-check-interval: Deprecated, use health-check-interval instead.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4.2-option%20httpchk
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-addr
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-port
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-inter
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-rise
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-fall

HSTS

Configure HSTS - HTTP Strict Transport Security. The following keys are supported:

• hsts: true if HSTS response header should be added
• hsts-include-subdomains: true if it should apply to subdomains as well
• hsts-max-age: time in seconds the browser should remember this configuration
• hsts-preload: true if the browser should include the domain to HSTS preload list

See also:

• https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Strict-Transport-Security

Initial weight

Configures the weight value of each backend server - either the enabled and also the disabled servers. The default value is 1. Changing this value has no effect on the proportional value between each server of a single backend, thus this doesn’t change the balance between the servers.

Change the default value to a higher number, eg 100, if using with agent-check and the agent is used to change the weight of the server.

Blue/green on deploy mode also uses initial-weight as its minimum weight value, provided that the maximum is lesser than or equal 256.

See also:

• agent-check
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-weight

Limit

Configure rate limit and concurrent connections per client IP address in order to mitigate DDoS attack. If several users are hidden behind the same IP (NAT or proxy), this configuration may have a negative impact for them. Whitelist can be used to these IPs.

The following annotations are supported:

• limit-connections: Maximum number os concurrent connections per client IP
• limit-rps: Maximum number of connections per second of the same IP
• limit-whitelist: Comma separated list of CIDRs that should be removed from the rate limit and concurrent connections check

Load server state

Define if HAProxy should save and reload it’s current state between server reloads, like uptime of backends, qty of requests and so on.

This is an experimental feature and has currently some issues if using with dynamic-scaling: an old state with disabled servers will disable them in the new configuration.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-server-state-file
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-load-server-state-from-file

Log format

Customize the tcp, http or https log format using log format variables. Only used if syslog-endpoint is also configured.

• http-log-format: log format of all HTTP proxies, defaults to HAProxy default HTTP log format.
• https-log-format: log format of TCP proxy used to inspect SNI extention. Use default to configure default TCP log format, defaults to not log.
• tcp-log-format: log format of TCP proxies, defaults to HAProxy default TCP log format. See also TCP services configmap command-line option.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#8.2.4
• syslog

Master-worker

Configures master-worker related options. These options are only used when an external haproxy instance is configured.

• master-exit-on-failure: If true, kill all the remaining workers and exit from master in the case of an unexpected failure of a worker, eg a segfault.
• worker-max-reloads: Defines how many reloads a haproxy worker should survive before receive a SIGTERM. The default value is 0 which means unlimited. This option limits the number of active workers and the haproxy’s pod memory usage. Useful on workloads with long running connections, eg websockets, and clusters that frequently changes and forces haproxy to reload.

See also:

• Example page
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-master-worker
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#mworker-max-reloads
• master-socket command-line option

Modsecurity

Configure modsecurity agent. These options only have effect if modsecurity-endpoints is configured.

Configure modsecurity-endpoints with a comma-separated list of IP:port of HAProxy agents (SPOA) for ModSecurity. The default configuration expects the contrib/modsecurity implementation from HAProxy source code.

Up to v0.7 all http requests will be parsed by the ModSecurity agent, even if the ingress resource wasn’t configured to deny requests based on ModSecurity response. Since v0.8 the spoe filter is configured on a per-backend basis.

The following keys are supported:

• modsecurity-endpoints: Comma separated list of ModSecurity agent endpoints.
• modsecurity-timeout-connect: Defines the maximum time to wait for the connection to the agent be established. Configures the haproxy’s timeout connect. Defaults to 5s if not configured.
• modsecurity-timeout-hello: Defines the maximum time to wait for the AGENT-HELLO frame from the agent. Default value is 100ms.
• modsecurity-timeout-idle: Defines the maximum time to wait before close an idle connection. Default value is 30s.
• modsecurity-timeout-processing: Defines the maximum time to wait for the whole ModSecurity processing. Default value is 1s.
• modsecurity-timeout-server: Defines the maximum time to wait for an agent response. Configures the haproxy’s timeout server. Defaults to 5s if not configured.

See also:

• example page.
• waf configuration key.
• https://www.haproxy.org/download/2.0/doc/SPOE.txt
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#9.3
• https://github.com/jcmoraisjr/modsecurity-spoa

Nbproc

Define the number of dedicated HAProxy process to the SSL/TLS handshake and offloading. The default value is 0 (zero) which means HAProxy should process all the SSL/TLS offloading, as well as the header inspection and load balancing within the same HAProxy process.

The recommended value depends on how much CPU a single HAProxy process is spending. Use 0 (zero) if the amount of processing has low CPU usage. This will avoid a more complex topology and an inter-process communication. Use the number of cores of a dedicated host minus 1 (one) to distribute the SSL/TLS offloading process. Leave one core dedicated to header inspection and load balancing.

If splitting HAProxy into two or more process and the number of threads is one, cpu-map is used to bind each process on its own CPU core.

See also:

• nbthread configuration key
• cpu-map configuration key
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-nbproc
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-bind-process
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-cpu-map

Nbthread

Define the number of threads a single HAProxy process should use to all its processing. If using with nbproc, every single HAProxy process will share this same configuration.

If using two or more threads on a single HAProxy process, cpu-map is used to bind each thread on its own CPU core.

See also:

• cpu-map configuration key
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-nbthread
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-cpu-map

OAuth

Configure OAuth2 via Bitly’s oauth2_proxy.

• oauth: Defines the oauth implementation. The only supported option is oauth2_proxy.
• oauth-uri-prefix: Defines the URI prefix of the oauth service. The default value is /oauth2. There should be a backend with this path in the ingress resource.
• oauth-headers: Defines an optional comma-separated list of <header>:<haproxy-var> used to configure request headers to the upstream backends. The default value is X-Auth-Request-Email:auth_response_email which means configuring a header X-Auth-Request-Email with the value of the var auth_response_email. New variables can be added overwriting the default auth-request.lua script.

The oauth2_proxy implementation expects Bitly’s oauth2_proxy running as a backend of the same domain that should be protected. oauth2_proxy has support to GitHub, Google, Facebook, OIDC and many others.

Note to v0.7 or below: All paths of a domain will have the same oauth configurations, despite if the path is configured on an ingress resource without oauth annotations. In other words, if two ingress resources share the same domain but only one has oauth annotations - the one that has at least the oauth2_proxy service - all paths from that domain will be protected.

See also:

• external-has-lua configuration key.
• example page.

Path type

Defines how the path of an incoming request should match a declared path in the ingress object.

• path-type: Configures the path type. Case insensitive, so Begin and begin configures the same path type option. The ingress spec has priority, this option will only be used if the pathType attribute from the ingress spec is declared as ImplementationSpecific.
• path-type-order: Defines a comma-separated list of the order that non overlapping paths should be matched. The default order starts with exact which is the fastest match, and finishes with regex which is the slowest one. All path types must be provided. Case insensitive, use all path types in lowercase. In the case of overlapping paths, eg /dir/sub and /dir, the one with the highest precedence, /dir/sub in the example, will always be checked first despite its match type.

Supported path-type values:

• begin: Case insensitive, matches the beginning of the path from the incoming request. This is the default value if not declared.
• exact: Case sensitive, matches the whole path. Implements the Exact path type from the ingress spec.
• prefix: Case sensitive, matches a whole subdirectory from the incoming path. A declared /app path matches /app and /app/1 but does not match /app1. Implements the Prefix path type from the ingress spec.
• regex: Case sensitive, matches the incoming path using POSIX extended regular expression. The regular expression has an implicit start ^ and no ending $ boundary, so a declared /app[0-9]+/? will match paths starting with this pattern. Add a trailing $ if an exact match is desired.

Request and match examples:

Proxy body size

Define the maximum number of bytes HAProxy will allow on the body of requests. Default is to not check, which means requests of unlimited size. This limit can be changed per ingress resource.

Since 0.4 a suffix can be added to the size, so 10m means 10 * 1024 * 1024 bytes. Supported suffix are: k, m and g.

Since 0.7 unlimited can also be used to overwrite any global body size limit.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#7.3.6-req.body_size

Proxy protocol

Configures PROXY protocol in frontends and backends.

• proxy-protocol: Define if the upstream backends support proxy protocol and what version of the protocol should be used. Supported values are v1, v2, v2-ssl, v2-ssl-cn or no. The default behavior if not declared is that the protocol is not supported by the backends and should not be used.
• use-proxy-protocol: Define if HAProxy is behind another proxy that use the PROXY protocol. If true, ports 80 and 443 will expect the PROXY protocol. The stats endpoint (defaults to port 1936) has it’s own stats-proxy-protocol configuration key.

See also:

• https://www.haproxy.org/download/2.0/doc/proxy-protocol.txt
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.1-accept-proxy
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-send-proxy
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-send-proxy-v2
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-send-proxy-v2-ssl
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-send-proxy-v2-ssl-cn

Rewrite target

Configures how URI of the requests should be rewritten before send the request to the backend. The following table shows some examples:

Secure backend

Configure secure (TLS) connection to the backends.

• secure-backends: Define as true if the backend provide a TLS connection.
• secure-crt-secret: Optional secret name of client certificate and key. This cert/key pair must be provided if the backend requests a client certificate. Expected secret keys are tls.crt and tls.key, the same used if secret is built with kubectl create secret tls <name>.
• secure-sni: Optional hostname that should be used as the SNI TLS extension sent to the backend server. If host is used as the content, the header Host from the incoming request is used as the SNI extension in the request to the backend. sni can also be used, which will use the same SNI from the incoming request. Note that, although the header Host is always right, the incoming SNI might be wrong if a TLS connection that’s already opened is reused - this is a common practice on browsers connecting over http2. Any other value different of host or sni will be used verbatim and should be a valid domain. If secure-verify-ca-secret is also provided, this hostname is also used to validate the server certificate names.
• secure-verify-ca-secret: Optional but recommended secret name with certificate authority bundle used to validate server certificate, preventing man-in-the-middle attacks. Expected secret key is ca.crt. Since v0.9, an optional ca.crl key can also provide a CRL in PEM format for the server to verify against. Configure either secure-sni or secure-verify-hostname to verify the certificate name.
• secure-verify-hostname: Optional hostname used to verify the name of the server certificate, without using the SNI TLS extension. This option can only be used if secure-verify-ca-secret was provided, and only supports harcoded domains which is used verbatim.

See also:

• Backend protocol configuration key.
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-verify
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-verifyhost
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-sni

Security

Change security options.

• username and groupname: Changes the user and group names used to run haproxy as non root. The default value is an empty string, which means leave haproxy running as root. Note that even running as root, haproxy always drops its own privileges before start its event loop. Both options should be declared to the configuration take effect. Note that this configuration means “running haproxy as non root”, it’s only useful when the haproxy container starts as root.
• use-chroot: If true, configures haproxy to perform a chroot() in the empty and non-writable directory /var/empty during the startup process, just before it drops its own privileges. Only root can perform a chroot(), so HAProxy Ingress container should start as UID 0 if this option is configured as true. See Using chroot() section below.
• use-haproxy-user: If true, configures username and groupname configuration keys as haproxy. See username and groupname above. Note that this user and group exists in the embedded haproxy, and should exist in the external haproxy if used. In the case of a conflict, username and groupname declaration will have priority and use-haproxy-user will be ignored. If false, the default value, user and group names will not be changed.

Starting as non root

In the default configuration HAProxy Ingress container starts as root. Since v0.9 it’s also possible to configure the container to start as haproxy user, UID 1001.

If using the embedded haproxy, read the Security considerations from HAProxy doc before change the starting user.

If using an external haproxy, configures the pod’s securityContext (instead of the container’s one) which will make Kubernetes create the shared file system with write access, so the controller can create and update configuration, maps and certificate files.

The starting user can be changed in the deployment or daemonset’s pod template using the following configuration:

...
  template:
    spec:
      securityContext:
        runAsUser: 1001

Note that ports below 1024 cannot be bound if the container starts as non-root.

Using chroot()

Beware of some chroot limitations:

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/management.html#13
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-chroot
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-uid
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-gid
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-unix-bind

Server alias

Configure hostname alias. All annotations will be combined together with the host attribute in the same ACL, and any of them might be used to match SNI extensions (TLS) or Host HTTP header. The matching is case insensitive.

• server-alias: Defines an alias with hostname-like syntax. On v0.6 and older, wildcard * wasn’t converted to match a subdomain. Regular expression was also accepted but dots were escaped, making this alias less useful as a regex. Starting v0.7 the same hostname syntax is used, so *.my.domain will match app.my.domain but won’t match sub.app.my.domain.
• server-alias-regex: Only in v0.7 and newer. Match hostname using a POSIX extended regular expression. The regex will be used verbatim, so add ^ and $ if strict hostname is desired and escape \. dots in order to strictly match them. Some HTTP clients add the port number in the Host header, so remember to add (:[0-9]+)?$ in the end of the regex if a dollar sign $ is being used to match the end of the string.

Service upstream

Defines if the HAProxy backend/server endpoints should be configured with the service VIP/IPVS. If false, the default value, the endpoints will be used and HAProxy will load balance the requests between them. If defined as true the service’s ClusterIP is used instead.

SSL always add HTTPS

Every hostname declared on an Ingress resource is added to an internal HTTP map. If at least one Ingress adds the hostname in the tls attribute, the hostname is also added to an internal HTTPS map and does ssl offload using the default certificate. A secret name can also be added in the tls attribute, overriding the certificate used in the TLS handshake.

ssl-always-add-https asks the controller to always add the domain in the internal HTTP and HTTPS maps, even if the tls attribute isn’t declared. If false, a missing tls attribute will only declare the domain in the HTTP map and ssl-redirect is ignored. If true, a missing tls attribute adds the domain in the HTTPS map, and the TLS handshake will use the default certificate. If tls attribute is used, this configuration is ignored.

The default value is true up to v0.12 releases to preserve historical behavior of this controller. The default value can be globally changed in the global ConfigMap.

SSL ciphers

Set the list of cipher algorithms used during the SSL/TLS handshake.

• ssl-cipher-suites: Cipher suites on TLS v1.3 handshake of incoming requests. HAProxy being the TLS server.
• ssl-cipher-suites-backend: Cipher suites on TLS v1.3 handshake to backend/servers. HAProxy being the TLS client.
• ssl-ciphers: Cipher suites on TLS up to v1.2 handshake of incoming requests. HAProxy being the TLS server.
• ssl-ciphers-backend: Cipher suites on TLS up to v1.2 handshake to backend/servers. HAProxy being the TLS client.

Default values on HAProxy Ingress up to v0.8:

• TLS up to v1.2: ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-AES256-GCM-SHA384:DHE-RSA-AES128-GCM-SHA256:DHE-DSS-AES128-GCM-SHA256:kEDH+AESGCM:ECDHE-RSA-AES128-SHA256:ECDHE-ECDSA-AES128-SHA256:ECDHE-RSA-AES128-SHA:ECDHE-ECDSA-AES128-SHA:ECDHE-RSA-AES256-SHA384:ECDHE-ECDSA-AES256-SHA384:ECDHE-RSA-AES256-SHA:ECDHE-ECDSA-AES256-SHA:DHE-RSA-AES128-SHA256:DHE-RSA-AES128-SHA:DHE-DSS-AES128-SHA256:DHE-RSA-AES256-SHA256:DHE-DSS-AES256-SHA:DHE-RSA-AES256-SHA:!aNULL:!eNULL:!EXPORT:!DES:!RC4:!3DES:!MD5:!PSK

Default values on HAProxy Ingress v0.9 and newer:

• TLS up to v1.2: ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-CHACHA20-POLY1305:ECDHE-RSA-CHACHA20-POLY1305:DHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES256-GCM-SHA384
• TLS v1.3: TLS_AES_128_GCM_SHA256:TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256

ssl-ciphers and ssl-cipher-suites were Global scope up to v0.10.

See also:

• https://ssl-config.mozilla.org/#server=haproxy
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-ssl-default-bind-ciphers
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-ssl-default-bind-ciphersuites
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-ciphers
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.2-ciphersuites

SSL DH

Configures Diffie-Hellman key exchange parameters.

• ssl-dh-param: Configure the secret name which defines the DH parameters file used on ephemeral Diffie-Hellman key exchange during the SSL/TLS handshake.
• ssl-dh-default-max-size: Define the maximum size of a temporary DH parameters used for key exchange. Only used if ssl-dh-param isn’t provided.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#tune.ssl.default-dh-param
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-ssl-dh-param-file

SSL engine

Set the name of the OpenSSL engine to use. The string shall include the engine name and its parameters.

Additionally, ssl-mode-async can be set to enable asynchronous TLS I/O operations if the ssl-engine used supports it.

Reference:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#ssl-engine
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#ssl-mode-async

SSL options

Define a space-separated list of options on SSL/TLS connections.

• ssl-options: Default options for all the TLS frontend connections - HAProxy being the server
• ssl-options-backend: Options for backend server connections - HAProxy being the client
• ssl-options-host: Options for TLS frontend connections - HAProxy being the server. This acts as a host scoped override to options defined in ssl-options and supports everything that HAProxy supports in the crt-list.

Default values for ssl-options and ssl-options-backend:

• v0.9 and newer: no-sslv3 no-tlsv10 no-tlsv11 no-tls-tickets
• up to v0.8: no-sslv3 no-tls-tickets

Supported options for ssl-options and ssl-options-backend:

• force-sslv3: Enforces use of SSLv3 only
• force-tlsv10: Enforces use of TLSv1.0 only
• force-tlsv11: Enforces use of TLSv1.1 only
• force-tlsv12: Enforces use of TLSv1.2 only
• no-sslv3: Disables support for SSLv3
• no-tls-tickets: Enforces the use of stateful session resumption
• no-tlsv10: Disables support for TLSv1.0
• no-tlsv11: Disables support for TLSv1.1
• no-tlsv12: Disables support for TLSv1.2

New supported options since v0.9 for ssl-options and ssl-options-backend:

• force-tlsv13: Enforces use of TLSv1.3 only
• no-tlsv13: Disables support for TLSv1.3
• ssl-max-ver <SSLv3|TLSv1.0|TLSv1.1|TLSv1.2|TLSv1.3>: Enforces the use of a SSL/TLS version or lower
• ssl-min-ver <SSLv3|TLSv1.0|TLSv1.1|TLSv1.2|TLSv1.3>: Enforces the use of a SSL/TLS version or upper

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.1-crt-list

SSL passthrough

Defines if HAProxy should work in TCP proxy mode and leave the SSL offload to the backend. SSL passthrough is a per domain configuration, which means that other domains can be configured to SSL offload on HAProxy.

If using SSL passthrough, only root / path is supported.

• ssl-passthrough: Enable SSL passthrough if defined as true. The backend is then expected to SSL offload the incoming traffic. The default value is false, which means HAProxy should do the SSL handshake.
• ssl-passthrough-http-port: Since v0.7. Optional HTTP port number of the backend. If defined, connections to the HAProxy HTTP port, default 80, is sent to that port which expects to speak plain HTTP. If not defined, connections to the HTTP port will redirect connections to the HTTPS one.

SSL redirect

Configures if an encripted connection should be used.

• ssl-redirect: Defines if HAProxy should send a 302 redirect response to requests made on unencripted connections. Note that this configuration will only make effect if TLS is configured.
• ssl-redirect-code: Defines the HTTP status code used in the redirect. The default value is 302 if not declared. Supported values are 301, 302, 303, 307 and 308.
• no-tls-redirect-locations: Defines a comma-separated list of URLs that should be removed from the TLS redirect. Requests to :80 http port and starting with one of the URLs from the list will not be redirected to https despite of the TLS redirect configuration. This option defaults to /.well-known/acme-challenge, used by ACME protocol.

See also:

• ssl-always-add-https configuration key
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#redirect

Stats

Configurations of the HAProxy statistics page:

• stats-auth: Enable basic authentication with clear-text password - <user>:<passwd>
• stats-port: Change the port HAProxy should listen to requests
• stats-proxy-protocol: Define if the stats endpoint should enforce the PROXY protocol
• stats-ssl-cert: Optional namespace/secret-name of tls.crt and tls.key pair used to enable SSL on stats page. Plain http will be used if not provided, the secret wasn’t found or the secret doesn’t have a crt/key pair.

Strict host

Defines whether the path of another matching host/FQDN should be used to try to serve a request. The default value is false, which means all matching wildcard hosts will be visited in order to try to match the path. If true, a strict configuration is applied and the default-backend should be used if a path couldn’t be matched.

Using the following configuration:

  spec:
    rules:
    - host: my.domain.com
      http:
        paths:
        - path: /a
          backend:
            serviceName: svc1
            servicePort: 8080
    - host: *.domain.com
      http:
        paths:
        - path: /
          backend:
            serviceName: svc2
            servicePort: 8080

A request to my.domain.com/b would serve:

• svc2 if strict-host is false, the default value
• default-backend if strict-host is true

Syslog

Logging configurations.

• syslog-endpoint: Configures the UDP syslog endpoint where HAProxy should send access logs.
• syslog-format: Configures the log format to be either rfc5424 (default), rfc3164 or raw.
• syslog-length: The maximum line length, log lines larger than this value will be truncated. Defaults to 1024.
• syslog-tag: Configure the tag field in the syslog header to the supplied string.

The HAProxy process can also send logs to stdout, instead of an external syslog endpoint or a syslog sidecar, by following the steps below:

• Configure syslog-endpoint as stdout and syslog-format as raw
• From v0.12 and newer, configure HAProxy to run as a sidecar, see the example page

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-log
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-log-tag

Timeout

Define timeout configurations. The unit defaults to milliseconds if missing, change the unit with s, m, h, … suffix.

The following keys are supported:

• timeout-client: Maximum inactivity time on the client side
• timeout-client-fin: Maximum inactivity time on the client side for half-closed connections - FIN_WAIT state
• timeout-connect: Maximum time to wait for a connection to a backend
• timeout-http-request: Maximum time to wait for a complete HTTP request
• timeout-keep-alive: Maximum time to wait for a new HTTP request on keep-alive connections
• timeout-queue: Maximum time a connection should wait on a server queue before return a 503 error to the client
• timeout-server: Maximum inactivity time on the backend side
• timeout-server-fin: Maximum inactivity time on the backend side for half-closed connections - FIN_WAIT state
• timeout-stop: Maximum time to wait for long lived connections to finish, eg websocket, before hard-stop a HAProxy process due to a reload
• timeout-tunnel: Maximum inactivity time on the client and backend side for tunnels

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#3.1-hard-stop-after (timeout-stop)
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#2.4 (time suffix)

TLS ALPN

Defines the TLS ALPN extension advertisement. The default value is h2,http/1.1 which enables HTTP/2 on the client side.

tls-alpn was Global scope up to v0.10.

See also:

• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#5.1-alpn

Use HTX

Defines if the new HTX internal representation for HTTP elements should be used. The default value is true since v0.10, it was false on v0.9. HTX should be used to enable HTTP/2 protocol to backends.

See also:

• backend-protocol configuration keys
• https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4-option%20http-use-htx

Var namespace

If var-namespace is configured as true, a HAProxy var txn.namespace is created with the kubernetes namespace owner of the service which is the target of the request. This variable is useful on http logs. The default value is false. Usage: k8s-namespace: %[var(txn.namespace)].

See also:

• http-log configuration key

WAF

Defines which web application firewall (WAF) implementation should be used to validate requests. Currently the only supported value is modsecurity.

This configuration has no effect if the ModSecurity endpoints are not configured.

The waf-mode key defines wether the WAF should be deny or detect for that Backend. If the WAF is in detect mode the requests are passed to ModSecurity and logged, but not denied.

The default behavior here is deny if waf is set to modsecurity.

See also:

• Modsecurity configuration keys.