Ingress configuration on Google Cloud


Overview

This page provides a comprehensive overview of what you can configure through Kubernetes Ingress on Google Cloud. The document also compares supported features for Ingress on Google Cloud and provides instructions for configuring Ingress using the default controller, FrontendConfig parameters, and BackendConfig parameters.

This page is for Networking specialists who design and architect the network for their organization and install, configure, and support network equipment. To learn more about common roles and example tasks that we reference in Google Cloud content, see Common GKE Enterprise user roles and tasks.

Feature comparison

The following table provides a list of supported features for Ingress on Google Cloud. The availability of the feature, General availability (GA) or Beta is also indicated.

Ingress class External Ingress Internal Ingress Multi Cluster Ingress
Ingress controller Google-hosted Ingress controller
Google Cloud load balancer type External HTTP(S) LB Internal HTTP(S) LB External HTTP(S) LB
Cluster scope Single cluster Single cluster Multi-cluster
Load balancer scope Global Regional Global
Environment support GKE GKE GKE
Shared VPC support GA GA GA
Service annotations
Container-native Load Balancing (NEGs) GA GA GA
HTTPS from load balancer to backends GA GA GA
HTTP/2 GA GA
TLS Only
GA
Ingress annotations
Static IP addresses GA GA GA
Kubernetes Secrets-based certificates GA GA GA
Self-managed SSL certificates GA GA GA
Google-managed SSL certificates GA GA
FrontendConfig
SSL policy GA GA with Gateway GA
HTTP-to-HTTPS redirect GA
1.17.13-gke.2600+GA
GA
BackendConfig
Backend service timeout GA GA GA
Cloud CDN GA GA
Connection draining timeout GA GA GA
Custom load balancer health check configuration GA GA GA
Google Cloud Armor security policy GA
1.19.10-gke.700G
GA
HTTP access logging configuration GA GA GA
Identity-Aware Proxy (IAP) GA GA GA
Session affinity GA GA GA
User-defined request headers GA GA
Custom response headers GA
1.25-gke+G
GA
1.25-gke+G

BThis feature is available in beta starting from the specified version. Features without a version listed are supported for all available GKE versions.

GThis feature is supported as GA starting from the specified version.

Configuring Ingress using the default controller

You cannot manually configure LoadBalancer features using the Google Cloud SDK or the Google Cloud console. You must use BackendConfig or FrontendConfig Kubernetes resources.

When creating an Ingress using the default controller, you can choose the type of load balancer (an external Application Load Balancer or an internal Application Load Balancer) by using an annotation on the Ingress object. You can choose whether GKE creates zonal NEGs or if it uses instance groups by using an annotation on each Service object.

FrontendConfig and BackendConfig custom resource definitions (CRDs) allow you to further customize the load balancer. These CRDs allow you to define additional load balancer features hierarchically, in a more structured way than annotations. To use Ingress (and these CRDs), you must have the HTTP load balancing add-on enabled. GKE clusters have HTTP load balancing enabled by default; you must not disable it.

FrontendConfigs are referenced in an Ingress object and can only be used with external Ingresses. BackendConfigs are referenced by a Service object. The same CRDs can be referenced by multiple Service or Ingress objects for configuration consistency. The FrontendConfig and BackendConfig CRDs share the same lifecycle as their corresponding Ingress and Service resources and they are often deployed together.

The following diagram illustrates how:

  • An annotation on an Ingress or MultiClusterIngress object references a FrontendConfig CRD. The FrontendConfig CRD references a Google Cloud SSL Policy.

  • An annotation on a Service or MultiClusterService object references a BackendConfig CRD. The BackendConfig CRD specifies custom settings for the corresponding backend service's health check.

BackendConfig and FrontendConfig overview
Figure: BackendConfig and FrontendConfig overview

Associating FrontendConfig with your Ingress

FrontendConfig can only be used with External Ingresses.

You can associate a FrontendConfig with an Ingress or a MultiClusterIngress.

Ingress

Use the networking.gke.io/v1beta1.FrontendConfig annotation:

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  annotations:
    networking.gke.io/v1beta1.FrontendConfig: "FRONTENDCONFIG_NAME"
...

Replace FRONTENDCONFIG_NAME with the name of your FrontendConfig.

MultiClusterIngress

Use the networking.gke.io/frontend-config annotation:

apiVersion: networking.gke.io/v1
kind: MultiClusterIngress
metadata:
  annotations:
    networking.gke.io/frontend-config: "FRONTENDCONFIG_NAME"
...

Replace FRONTENDCONFIG_NAME with the name of your FrontendConfig.

Associating BackendConfig with your Ingress

You can use the cloud.google.com/backend-config or beta.cloud.google.com/backend-config annotation to specify the name of a BackendConfig.

Same BackendConfig for all Service ports

To use the same BackendConfig for all ports, use the default key in the annotation. The Ingress controller uses the same BackendConfig each time it creates a load balancer backend service to reference one of the Service's ports.

You can use the default key for both Ingress and MultiClusterIngress resources.
apiVersion: v1
kind: Service
metadata:
  annotations:
    cloud.google.com/backend-config: '{"default": "my-backendconfig"}'
...

Unique BackendConfig per Service port

For both Ingress and MultiClusterIngress, you can specify a custom BackendConfig for one or more ports using a key that matches the port's name or number. The Ingress controller uses the specific BackendConfig when it creates a load balancer backend service for a referenced Service port.

GKE 1.16-gke.3 and later

apiVersion: v1
kind: Service
metadata:
  annotations:
    cloud.google.com/backend-config: '{"ports": {
    "SERVICE_REFERENCE_A":"BACKENDCONFIG_REFERENCE_A",
    "SERVICE_REFERENCE_B":"BACKENDCONFIG_REFERENCE_B"
    }}'
spec:
  ports:
  - name: PORT_NAME_1
    port: PORT_NUMBER_1
    protocol: TCP
    targetPort: 50000
  - name: PORT_NAME_2
    port: PORT_NUMBER_2
    protocol: TCP
    targetPort: 8080
...

All supported versions

apiVersion: v1
kind: Service
metadata:
  annotations:
    cloud.google.com/backend-config: '{"ports": {
      PORT_NAME_1:"BACKENDCONFIG_REFERENCE_A",
      PORT_NAME_2:"BACKENDCONFIG_REFERENCE_B"
    }}'
spec:
  ports:
  - name: PORT_NAME_1
    port: PORT_NUMBER_1
    protocol: TCP
    targetPort: 50000
  - name: PORT_NAME_2
    port: PORT_NUMBER_2
    protocol: TCP
    targetPort: 8080
...

Replace the following:

  • BACKENDCONFIG_REFERENCE_A: the name of an existing BackendConfig.
  • BACKENDCONFIG_REFERENCE_B: the name of an existing BackendConfig.
  • SERVICE_REFERENCE_A: use the value of PORT_NUMBER_1 or PORT_NAME_1. This is because a Service's BackendConfig annotation can refer to either the port's name (spec.ports[].name) or the port's number (spec.ports[].port).
  • SERVICE_REFERENCE_B: use the value of PORT_NUMBER_2 or PORT_NAME_2. This is because a Service's BackendConfig annotation can refer to either the port's name (spec.ports[].name) or the port's number (spec.ports[].port).

When referring to the Service's port by number, you must use the port value instead of the targetPort value. The port number used here is only for binding the BackendConfig; it does not control the port to which the load balancer sends traffic or health check probes:

  • When using container-native load balancing, the load balancer sends traffic to an endpoint in a network endpoint group (matching a Pod IP address) on the referenced Service port's targetPort (which must match a containerPort for a serving Pod). Otherwise, the load balancer sends traffic to a node's IP address on the referenced Service port's nodePort.

  • When using a BackendConfig to provide a custom load balancer health check, the port number you use for the load balancer's health check can differ from the Service's spec.ports[].port number. For details about port numbers for health checks, see Custom health check configuration.

Configuring Ingress features through FrontendConfig parameters

The following section shows you how to set your FrontendConfig to enable specific Ingress features.

SSL policies

SSL policies allow you to specify a set of TLS versions and ciphers that the load balancer uses to terminate HTTPS traffic from clients. You must first create an SSL policy outside of GKE. Once created, you can reference it in a FrontendConfig CRD.

The sslPolicy field in the FrontendConfig references the name of an SSL policy in the same Google Cloud project as the GKE cluster. It attaches the SSL policy to the target HTTPS proxy, which was created for the external HTTP(S) load balancer by the Ingress. The same FrontendConfig resource and SSL policy can be referenced by multiple Ingress resources. If a referenced SSL policy is changed, the change is propagated to the Google Front Ends (GFEs) that power your external HTTP(S) load balancer created by the Ingress.

The following FrontendConfig manifest enables an SSL policy named gke-ingress-ssl-policy:

apiVersion: networking.gke.io/v1beta1
kind: FrontendConfig
metadata:
  name: my-frontend-config
spec:
  sslPolicy: gke-ingress-ssl-policy

HTTP to HTTPS redirects

An external HTTP load balancer can redirect unencrypted HTTP requests to an HTTPS load balancer that uses the same IP address. When you create an Ingress with HTTP to HTTPS redirects enabled, both of these load balancers are created automatically. Requests to the external IP address of the Ingress on port 80 are automatically redirected to the same external IP address on port 443. This functionality is built on HTTP to HTTPS redirects provided by Cloud Load Balancing.

To support HTTP to HTTPS redirection, an Ingress must be configured to serve both HTTP and HTTPS traffic. If either HTTP or HTTPS is disabled, redirection will not work.

HTTP to HTTPS redirects are configured using the redirectToHttps field in a FrontendConfig custom resource. Redirects are enabled for the entire Ingress resource so all services referenced by the Ingress will have HTTPS redirects enabled.

The following FrontendConfig manifest enables HTTP to HTTPS redirects. Set the spec.redirectToHttps.enabled field to true to enable HTTPS redirects. The spec.responseCodeName field is optional. If it's omitted a 301 Moved Permanently redirect is used.

apiVersion: networking.gke.io/v1beta1
kind: FrontendConfig
metadata:
  name: my-frontend-config
spec:
  redirectToHttps:
    enabled: true
    responseCodeName: RESPONSE_CODE

Replace RESPONSE_CODE with one of the following:

  • MOVED_PERMANENTLY_DEFAULT to return a 301 redirect response code (default if responseCodeName is unspecified).
  • FOUND to return a 302 redirect response code.
  • SEE_OTHER to return a 303 redirect response code.
  • TEMPORARY_REDIRECT to return a 307 redirect response code.
  • PERMANENT_REDIRECT to return a 308 redirect response code.

When redirects are enabled the Ingress controller creates a load balancer as shown in the following diagram:

A redirection-only external HTTP load balancer consisting of a forwarding rule, target HTTP proxy, and a URL map with a redirect to a complete HTTPS load balancer with backend services

To validate that your redirect is working, use a curl command:

curl http://IP_ADDRESS

Replace IP_ADDRESS with the IP address of your Ingress.

The response shows the redirect response code that you configured. For example the following example is for a FrontendConfig configured with a 301: MovedPermanently redirect:

<HTML><HEAD><meta http-equiv="content-type" content="text/html;charset=utf-8">
<TITLE>301 Moved</TITLE></HEAD><BODY>
<H1>301 Moved</H1>
The document has moved
<A HREF="https://35.244.160.59/">here</A>.</BODY></HTML>

Configuring Ingress features through BackendConfig parameters

The following sections show you how to set your BackendConfig to enable specific Ingress features. Changes to a BackendConfig resource are constantly reconciled, so you do not need to delete and recreate your Ingress to see that BackendConfig changes are reflected.

For information on BackendConfig limitations, see the limitations section.

Backend service timeout

You can use a BackendConfig to set a backend service timeout period in seconds. If you do not specify a value, the default value is 30 seconds.

The following BackendConfig manifest specifies a timeout of 40 seconds:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  timeoutSec: 40

Cloud CDN

You can enable Cloud CDN using a BackendConfig.

The following BackendConfig manifest shows all the fields available when enabling Cloud CDN:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  cdn:
    enabled: CDN_ENABLED
    cachePolicy:
      includeHost: INCLUDE_HOST
      includeProtocol: INCLUDE_PROTOCOL
      includeQueryString: INCLUDE_QUERY_STRING
      queryStringBlacklist: QUERY_STRING_DENYLIST
      queryStringWhitelist: QUERY_STRING_ALLOWLIST
    cacheMode: CACHE_MODE
    clientTtl: CLIENT_TTL
    defaultTtl: DEFAULT_TTL
    maxTtl: MAX_TTL
    negativeCaching: NEGATIVE_CACHING
    negativeCachingPolicy:
      code: NEGATIVE_CACHING_CODE
      ttl: NEGATIVE_CACHING_TTL
    requestCoalescing: REQ_COALESCING
    serveWhileStale: SERVE_WHILE_STALE
    signedUrlCacheMaxAgeSec: SIGNED_MAX_AGE
    signedUrlKeys:
      keyName: KEY_NAME
      keyValue: KEY_VALUE
      secretName: SECRET_NAME

Replace the following:

  • CDN_ENABLED: If set to true, Cloud CDN is enabled for this Ingress backend.
  • INCLUDE_HOST: If set to true, requests to different hosts are cached separately.
  • INCLUDE_PROTOCOL: If set to true, HTTP and HTTPS requests are cached separately.
  • INCLUDE_QUERY_STRING: If set to true, query string parameters are included in the cache key according to queryStringBlacklist or queryStringWhitelist. If neither is set, the entire query string is included. If set to false, the entire query string is excluded from the cache key.
  • QUERY_STRING_DENYLIST: Specify a string array with the names of query string parameters to exclude from cache keys. All other parameters are included. You can specify queryStringBlacklist or queryStringWhitelist, but not both.
  • QUERY_STRING_ALLOWLIST: Specify a string array with the names of query string parameters to include in cache keys. All other parameters are excluded. You can queryStringBlacklist or queryStringWhitelist, but not both.

The following fields are only supported in GKE versions 1.23.3-gke.900 and later using GKE Ingress. They are not supported using Multi Cluster Ingress:

Expand the following section to see an example that deploys Cloud CDN through Ingress and then validates that application content is being cached.

Connection draining timeout

You can configure connection draining timeout using a BackendConfig. Connection draining timeout is the time, in seconds, to wait for connections to drain. For the specified duration of the timeout, existing requests to the removed backend are given time to complete. The load balancer does not send new requests to the removed backend. After the timeout duration is reached, all remaining connections to the backend are closed. The timeout duration can be from 0 to 3600 seconds. The default value is 0, which also disables connection draining.

The following BackendConfig manifest specifies a connection draining timeout of 60 seconds:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  connectionDraining:
    drainingTimeoutSec: 60

Custom health check configuration

There are a variety of ways that GKE configures Google Cloud load balancer health checks when deploying through Ingress. To learn more about how GKE Ingress deploys health checks, see Ingress health checks.

A BackendConfig allows you to precisely control the load balancer health check settings.

You can configure multiple GKE Services to reference the same BackendConfig as a reusable template. For healthCheck parameters, a unique Google Cloud health check is created for each backend service corresponding to each GKE Service.

The following BackendConfig manifest shows all the fields available when configuring a BackendConfig health check:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  healthCheck:
    checkIntervalSec: INTERVAL
    timeoutSec: TIMEOUT
    healthyThreshold: HEALTH_THRESHOLD
    unhealthyThreshold: UNHEALTHY_THRESHOLD
    type: PROTOCOL
    requestPath: PATH
    port: PORT

Replace the following:

  • INTERVAL: Specify the check-interval, in seconds, for each health check prober. This is the time from the start of one prober's check to the start of its next check. If you omit this parameter, the Google Cloud default of 5 seconds is used. For additional implementation details, see Multiple probes and frequency.
  • TIMEOUT: Specify the amount of time that Google Cloud waits for a response to a probe. The value of TIMEOUT must be less than or equal to the INTERVAL. Units are seconds. Each probe requires an HTTP 200 (OK) response code to be delivered before the probe timeout.
  • HEALTH_THRESHOLD and UNHEALTHY_THRESHOLD: Specify the number of sequential connection attempts that must succeed or fail, for at least one prober, in order to change the health state from healthy to unhealthy or vice versa. If you omit one of these parameters, Google Cloud uses the default value of 2.
  • PROTOCOL: Specify a protocol used by probe systems for health checking. The BackendConfig only supports creating health checks using the HTTP, HTTPS, or HTTP2 protocols. For more information, see Success criteria for HTTP, HTTPS, and HTTP/2. You cannot omit this parameter.
  • PATH: For HTTP, HTTPS, or HTTP2 health checks, specify the request-path to which the probe system should connect. If you omit this parameter, Google Cloud uses the default of /.
  • PORT: A BackendConfig only supports specifying the load balancer health check port by using a port number. If you omit this parameter, Google Cloud uses the default value 80.

    • When using container native load balancing, you should select a port matching a containerPort of a serving Pod (whether or not that containerPort is referenced by a targetPort of the Service). Because the load balancer sends probes to the Pod's IP address directly, you are not limited to containerPorts referenced by a Service's targetPort. Health check probe systems connect to a serving Pod's IP address on the port you specify.

    • For instance group backends, you must select a port matching a nodePort exposed by the Service. Health check probe systems then connect to each node on that port.

To set up GKE Ingress with a custom HTTP health check, see GKE Ingress with custom HTTP health check.

Google Cloud Armor Ingress security policy

Google Cloud Armor security policies help you protect your load-balanced applications from web-based attacks. Once you have configured a Google Cloud Armor security policy, you can reference it using a BackendConfig.

Add the name of your security policy to the BackendConfig. The following BackendConfig manifest specifies a security policy named example-security-policy:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  namespace: cloud-armor-how-to
  name: my-backendconfig
spec:
  securityPolicy:
    name: "example-security-policy"

Two Sources Of Truth

Though configured through GKE, the underlying Compute Engine BackendService APIs can still be used to directly modify which security policy to apply. This creates two sources of truth: GKE and Compute Engine. GKE Ingress Controller's behavior in response to the securityPolicy field within BackendConfig is documented in the table below. To avoid conflict and unexpected behavior, we recommend using the GKE BackendConfig for the management of which security policy to use.

BackendConfig field Value Behaviour
spec.securityPolicy.name CloudArmorPolicyName The GKE Ingress Controller sets Google Cloud Armor policy named CloudArmorPolicyName to the load balancer. The GKE Ingress Controller overwrites whatever policy that was previously set.
spec.securityPolicy.name Empty string ("") The GKE Ingress Controller removes any configured Google Cloud Armor policy from the load balancer.
spec.securityPolicy nil The GKE Ingress Controller uses the configuration set on the BackendService object configured through the Compute Engine API using the Google Cloud console, or gcloud CLI, or Terraform.

To set up GKE Ingress with Google Cloud Armor protection, see Google Cloud Armor enabled Ingress.

HTTP access logging

Ingress can log all HTTP requests from clients to Cloud Logging. You can enable and disable access logging using BackendConfig along with setting the access logging sampling rate.

To configure access logging, use the logging field in your BackendConfig. If logging is omitted, access logging takes the default behavior. This is dependent on the GKE version.

You can configure the following fields:

  • enable: If set to true, access logging will be enabled for this Ingress and logs is available in Cloud Logging. Otherwise, access logging is disabled for this Ingress.
  • sampleRate: Specify a value from 0.0 through 1.0, where 0.0 means no packets are logged and 1.0 means 100% of packets are logged. This field is only relevant if enable is set to true. sampleRate is an optional field, but if it's configured then enable: true must also be set or else it is interpreted as enable: false.

The following BackendConfig manifest enables access logging and sets the sample rate to 50% of the HTTP requests for a given Ingress resource:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  logging:
    enable: true
    sampleRate: 0.5

Identity-Aware Proxy

To configure the BackendConfig for Identity-Aware Proxy IAP, you need to specify the enabled and secretName values to your iap block in your BackendConfig. To specify these values, ensure that you have the compute.backendServices.update permission.

The following BackendConfig manifest enables Identity-Aware Proxy:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name:  my-backendconfig
spec:
  iap:
    enabled: true
    oauthclientCredentials:
      secretName: my-secret

Enable IAP With Google Managed OAuth Client

Starting in GKE 1.29.4-gke.1043000, IAP can be configured to use Google Managed OAuth Client using a BackendConfig. To decide whether to use the Google OAuth Client or a custom OAuth Client, see When to use a custom OAuth configuration in the IAP documentation.

To enable IAP with Google Managed OAuth Client, do not provide the OAuthCredentials in the BackendConfig. For users who already configured IAP using OAuthCredentials there is no migration path to switch to using Google Managed OAuth Client, you must recreate the Backend (remove the Service from the Ingress and re-attach). We suggest performing this operation during a maintenance window as this will cause downtime. The opposite migration path, switching from Google Managed OAuthClient to OAuthCredentials is possible.

The following BackendConfig manifest enables Identity-Aware Proxy with Google Managed OAuth Client:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name:  my-backendconfig
spec:
  iap:
    enabled: true

For full instructions, see Enabling IAP for GKE in the IAP documentation.

Identity-Aware Proxy with internal Ingress

To configure internal Ingress for IAP, you must use the Premium Tier. If you do not use the Premium Tier, you cannot view or create internal Application Load Balancers with Identity-Aware Proxy. You must also have a Chrome Enterprise Premium subscription to use internal Ingress for IAP.

To set up secure GKE Ingress with Identity-Aware Proxy based authentication, see example, IAP enabled ingress.

Session affinity

You can use a BackendConfig to set session affinity to client IP or generated cookie.

Client IP affinity

To use a BackendConfig to set client IP affinity, set affinityType to "CLIENT_IP", as in the following example:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  sessionAffinity:
    affinityType: "CLIENT_IP"

To use a BackendConfig to set generated cookie affinity , set affinityType to GENERATED_COOKIE in your BackendConfig manifest. You can also use affinityCookieTtlSec to set the time period for the cookie to remain active.

The following manifest sets the affinity type to generated cookie and gives the cookies a TTL of 50 seconds:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  sessionAffinity:
    affinityType: "GENERATED_COOKIE"
    affinityCookieTtlSec: 50

User-defined request headers

You can use a BackendConfig to configure user-defined request headers. The load balancer adds the headers you specify to the requests it forwards to the backends.

The load balancer adds custom request headers only to the client requests, not to the health check probes. If your backend requires a specific header for authorization that is missing from the health check packet, the health check might fail.

To enable user-defined request headers, you specify a list of headers in the customRequestHeaders property of the BackendConfig resource. Specify each header as a header-name:header-value string.

The following BackendConfig manifest adds three request headers:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  customRequestHeaders:
    headers:
    - "X-Client-Region:{client_region}"
    - "X-Client-City:{client_city}"
    - "X-Client-CityLatLong:{client_city_lat_long}"

Custom response headers

To enable custom response headers, you specify a list of headers in the customResponseHeaders property of the BackendConfig resource. Specify each header as a header-name:header-value string.

Custom response headers are available only in GKE clusters version 1.25 and later.

The following BackendConfig manifest is an example to add an HTTP Strict Transport Security (HSTS) response header:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  customResponseHeaders:
    headers:
    - "Strict-Transport-Security: max-age=28800; includeSubDomains"

Exercise: Setting Ingress timeouts using a backend service

The following exercise shows you the steps required for configuring timeouts and connection draining with an Ingress with a BackendConfig resource.

To configure the backend properties of an Ingress, complete the following tasks:

  1. Create a Deployment.
  2. Create a BackendConfig.
  3. Create a Service, and associate one of its ports with the BackendConfig.
  4. Create an Ingress, and associate the Ingress with the (Service, port) pair.
  5. Validate the properties of the backend service.
  6. Clean up.

Creating a Deployment

Before you create a BackendConfig and a Service, you need to create a Deployment.

The following example manifest is for a Deployment named my-deployment.yaml:

# my-deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-deployment
spec:
  selector:
    matchLabels:
      purpose: bsc-config-demo
  replicas: 2
  template:
    metadata:
      labels:
        purpose: bsc-config-demo
    spec:
      containers:
      - name: hello-app-container
        image: us-docker.pkg.dev/google-samples/containers/gke/hello-app:1.0

Create the Deployment by running the following command:

kubectl apply -f my-deployment.yaml

Creating a BackendConfig

Use your BackendConfig to specify the Ingress features you want to use.

This BackendConfig manifest, named my-backendconfig.yaml, specifies:

  • A timeout of 40 seconds.
  • A connection draining timeout of 60 seconds.
# my-backendconfig.yaml
apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  timeoutSec: 40
  connectionDraining:
    drainingTimeoutSec: 60

Create the BackendConfig by running the following command:

kubectl apply -f my-backendconfig.yaml

Creating a Service

A BackendConfig corresponds to a single Service-Port combination, even if a Service has multiple ports. Each port can be associated with a single BackendConfig CRD. If a Service port is referenced by an Ingress, and if the Service port is associated with a BackendConfig, then the HTTP(S) load balancing backend service takes part of its configuration from the BackendConfig.

The following is an example Service manifest named my-service.yaml:

# my-service.yaml
apiVersion: v1
kind: Service
metadata:
  name: my-service
  labels:
    purpose: bsc-config-demo
  annotations:
    cloud.google.com/backend-config: '{"ports": {"80":"my-backendconfig"}}'
    cloud.google.com/neg: '{"ingress": true}'
spec:
  type: ClusterIP
  selector:
    purpose: bsc-config-demo
  ports:
  - port: 80
    protocol: TCP
    targetPort: 8080

The cloud.google.com/backend-config annotation specifies a mapping between ports and BackendConfig objects. In my-service.yaml:

  • Any Pod that has the label purpose: bsc-config-demo is a member of the Service.
  • TCP port 80 of the Service is associated with a BackendConfig named my-backendconfig. The cloud.google.com/backend-config annotation specifies this.
  • A request sent to port 80 of the Service is forwarded to one of the member Pods on port 8080.

To create the Service, run the following command:

kubectl apply -f my-service.yaml

Creating an Ingress

The following is an Ingress manifest named my-ingress.yaml. In this example, incoming requests are routed to port 80 of the Service named my-service.

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: my-ingress
spec:
  rules:
  - http:
      paths:
      - path: /*
        pathType: ImplementationSpecific
        backend:
          service:
            name: my-service
            port:
              number: 80

To create the Ingress, run the following command:

kubectl apply -f my-ingress.yaml

Wait a few minutes for the Ingress controller to configure an external Application Load Balancer and an associated backend service. Once this is complete, you have configured your Ingress to use a timeout of 40 seconds and a connection draining timeout of 60 seconds.

Validating backend service properties

You can validate that the correct load balancer settings have been applied through your BackendConfig. To do this, identify the backend service that Ingress has deployed and inspect its settings to validate that they match the Deployment manifests.

First, describe the my-ingress resource and filter for the annotation that lists the backend services associated with the Ingress. For example:

kubectl describe ingress my-ingress | grep ingress.kubernetes.io/backends

You should see output similar to the following:

ingress.kubernetes.io/backends: '{"k8s1-27fde173-default-my-service-80-8d4ca500":"HEALTHY","k8s1-27fde173-kube-system-default-http-backend-80-18dfe76c":"HEALTHY"}

The output provides information about your backend services. For example, this annotation contains two backend services:

  • "k8s1-27fde173-default-my-service-80-8d4ca500":"HEALTHY" provides information about the backend service associated with the my-service Kubernetes Service.
    • k8s1-27fde173 is a hash used to describe the cluster.
    • default is the Kubernetes namespace.
    • HEALTHY indicates that the backend is healthy.
  • "k8s1-27fde173-kube-system-default-http-backend-80-18dfe76c":"HEALTHY" provides information about the backend service associated with the default backend (404-server).
    • k8s1-27fde173 is a hash used to describe the cluster.
    • kube-system is the namespace.
    • default-http-backend is the Kubernetes Service name.
    • 80 is the host port.
    • HEALTHY indicates that the backend is healthy.

Next, inspect the backend service associated with my-service using gcloud. Filter for "drainingTimeoutSec" and "timeoutSec" to confirm that they've been set in the Google Cloud Load Balancer control plane. For example:

# Optionally, set a variable
export BES=k8s1-27fde173-default-my-service-80-8d4ca500

# Filter for drainingTimeoutSec and timeoutSec
gcloud compute backend-services describe ${BES} --global | grep -e "drainingTimeoutSec" -e "timeoutSec"

Output:

  drainingTimeoutSec: 60
  timeoutSec: 40

Seeing drainingTimeoutSec and timeoutSec in the output confirms that their values were correctly set through the BackendConfig.

Cleaning up

To prevent unwanted charges incurring on your account, delete the Kubernetes objects that you created for this exercise:

kubectl delete ingress my-ingress
kubectl delete service my-service
kubectl delete backendconfig my-backendconfig
kubectl delete deployment my-deployment

BackendConfig limitations

BackendConfigs have the following limitations:

  • Only one (Service, port) pair can consume only one BackendConfig, even if multiple Ingress objects reference the (Service, port). This means all Ingress objects that reference the same (Service, port) must use the same configuration for Google Cloud Armor, IAP, and Cloud CDN.
  • IAP and Cloud CDN cannot be enabled for the same HTTP(S) Load Balancing backend service. This means that you cannot configure both IAP and Cloud CDN in the same BackendConfig.
  • You must use kubectl 1.7 or later to interact with BackendConfig.

Removing the configuration specified in a FrontendConfig or BackendConfig

To revoke an Ingress feature, you must explicitly disable the feature configuration in the FrontendConfig or BackendConfig CRD. The Ingress controller only reconciles configurations specified in these CRDs.

To clear or disable a previously enabled configuration, set the field's value to an empty string ("") or to a Boolean value of false, depending on the field type.

The following BackendConfig manifest disables a Google Cloud Armor security policy and Cloud CDN:

apiVersion: cloud.google.com/v1
kind: BackendConfig
metadata:
  name: my-backendconfig
spec:
  cdn:
    enabled: false
  securityPolicy:
    name: ""

Deleting a FrontendConfig or BackendConfig

FrontendConfig

To delete a FrontendConfig, follow these steps:

  1. Remove the FrontendConfig's name from the networking.gke.io/v1beta1.FrontendConfig annotation in the Ingress manifest.

  2. Apply the changed Ingress manifest to your cluster. For example, use kubectl apply.

  3. Delete the FrontendConfig. For example, use kubectl delete frontendconfig config my-frontendconfig.

BackendConfig

To delete a BackedConfig, follow these steps:

  1. Remove the BackendConfig's name from the cloud.google.com/backend-config annotation in the Service manifest.

  2. Apply the changed Service manifest to your cluster. For example, use kubectl apply.

  3. Delete the BackendConfig. For example, use kubectl delete backendconfig my-backendconfig.

Troubleshooting

You can detect common misconfigurations using the Ingress diagnostic tool. You should also ensure that any health checks are configured correctly.

BackendConfig not found

This error occurs when a BackendConfig for a Service port is specified in the Service annotation, but the actual BackendConfig resource couldn't be found.

To evaluate a Kubernetes event, run the following command:

kubectl get event

The following example output indicates your BackendConfig was not found:

KIND    ... SOURCE
Ingress ... loadbalancer-controller

MESSAGE
Error during sync: error getting BackendConfig for port 80 on service "default/my-service":
no BackendConfig for service port exists

To resolve this issue, ensure you have not created the BackendConfig resource in the wrong namespace or misspelled its reference in the Service annotation.

Ingress security policy not found

After the Ingress object is created, if the security policy isn't properly associated with the LoadBalancer Service, evaluate the Kubernetes event to see if there is a configuration mistake. If your BackendConfig specifies a security policy that does not exist, a warning event is periodically emitted.

To evaluate a Kubernetes event, run the following command:

kubectl get event

The following example output indicates your security policy was not found:

KIND    ... SOURCE
Ingress ... loadbalancer-controller

MESSAGE
Error during sync: The given security policy "my-policy" does not exist.

To resolve this issue, specify the correct security policy name in your BackendConfig.

Addressing 500 series errors with NEGs during workload scaling in GKE

Symptom:

When you use GKE provisioned NEGs for load balancing, you might experience 502 or 503 errors for the services during the workload scale down. 502 errors occur when Pods are terminated before existing connections close, while the 503 errors occur when traffic is directed to deleted Pods.

This issue can affect clusters if you are using GKE managed load balancing products that use NEGs, including Gateway, Ingress, and standalone NEGs. If you frequently scale your workloads, your cluster is at a higher risk of being affected.

Diagnosis:

Removing a Pod in Kubernetes without draining its endpoint and removing it from the NEG first leads to 500 series errors. To avoid issues during Pod termination, you must consider the order of operations. The following images display scenarios when BackendService Drain Timeout is unset and BackendService Drain Timeout is set with BackendConfig.

Scenario 1: BackendService Drain Timeout is unset.

The following image displays a scenario where the BackendService Drain Timeout is unset.

BackendService Drain Timeout is unset.

Scenario 2: BackendService Drain Timeout is set.

The following image displays a scenario where the BackendService Drain Timeout is set.

BackendService Drain Timeout is set.

The exact time the 500 series errors occur depends on the following factors:

  • NEG API detach latency: The NEG API detach latency represents the current time taken for the detach operation to finalize in Google Cloud. This is influenced by a variety of factors outside Kubernetes, including the type of load balancer and the specific zone.

  • Drain latency: Drain latency represents the time taken for the load balancer to start directing traffic away from a particular part of your system. Once drain is initiated, the load balancer stops sending new requests to the endpoint, however there is still a latency in triggering drain (Drain Latency) which can cause temporary 503 errors if the Pod no longer exists.

  • Health check configuration: More sensitive health check thresholds mitigate the duration of 503 errors as it can signal the load balancer to stop sending requests to endpoints even if the detach operation has not finished.

  • Termination grace period: The termination grace period determines the maximum amount of time a Pod is given to exit. However, a Pod can exit before the termination grace period completes. If a Pod takes longer than this period, the Pod is forced to exit at the end of this period. This is a setting on the Pod and needs to be configured in the workload definition.

Potential resolution:

To prevent those 5XX errors, apply the following settings. The timeout values are suggestive and you might need to adjust them for your specific application. The following section guides you through the customization process.

The following image displays how to keep the Pod alive with preStopHook:

BackendService Drain Timeout is set.

To avoid 500 series errors, perform the following steps:

  1. Set the BackendService Drain Timeout for your service to 1 minute.

  2. Extend the terminationGracePeriod on the Pod.

    Set the terminationGracePeriodSeconds on the Pod to 3.5 minutes. When combined with the recommended settings, this allows your Pods a 30 to 45 second window for a graceful shutdown after the Pod's endpoint has been removed from the NEG. If you require more time for the graceful shutdown, you can extend the grace period or follow the instructions mentioned in the Customize timeouts section.

    The following Pod manifest specifies a connection draining timeout of 210 seconds (3.5 minutes):

    spec:
      terminationGracePeriodSeconds: 210
      containers:
      - name: my-app
        ...
      ...
    
  3. Apply a preStopHook to all containers.

    Apply a preStopHook that will ensure the Pod is alive for 120 seconds longer while the Pod's endpoint is drained in the load balancer and the endpoint is removed from the NEG.

    spec:
      containers:
      - name: my-app
        ...
        lifecycle:
          preStopHook:
            exec:
              command: ["/bin/sh", "-c", "sleep 120s"]
      ...
    

Customize timeouts

To ensure Pod continuity and prevent 500 series errors, the Pod must be alive until the endpoint is removed from the NEG. Specially to prevent 502 and 503 errors, consider implementing a combination of timeouts and a preStopHook.

To keep the Pod alive longer during the shutdown process, add a preStopHook to the Pod. Run the preStopHook before a Pod is signaled to exit, so the preStopHook can be used to keep the Pod around until its corresponding endpoint is removed from the NEG.

To extend the duration that a Pod remains active during the shutdown process, insert a preStopHook into the Pod configuration as follows:

spec:
  containers:
  - name: my-app
    ...
    lifecycle:
      preStopHook:
        exec:
          command: ["/bin/sh", "-c", "sleep <latency time>"]

You can configure timeouts and related settings to manage the graceful shutdown of Pods during workload scale downs. You can adjust timeouts based on specific use cases. We recommend that you start with longer timeouts and reduce the duration as necessary. You can customize the timeouts by configuring timeout-related parameters and the preStopHook in the following ways:

Backend Service Drain Timeout

The Backend Service Drain Timeout parameter is unset by default and has no effect. If you set the Backend Service Drain Timeout parameter and activate it, the load balancer stops routing new requests to the endpoint and waits the timeout before terminating existing connections.

You can set the Backend Service Drain Timeout parameter by using the BackendConfig with Ingress, the GCPBackendPolicy with Gateway or manually on the BackendService with standalone NEGs. The timeout should be 1.5 to 2 times longer than the time it takes to process a request. This ensures if a request came in right before the drain was initiated, it will complete before the timeout completes. Setting the Backend Service Drain Timeout parameter to a value greater than 0 helps mitigate 503 errors because no new requests are sent to endpoints scheduled for removal. For this timeout to be effective, you must use it in conjunction with the preStopHook to ensure that the Pod remains active while the drain occurs. Without this combination, existing requests that didn't complete will receive a 502 error.

preStopHook time

The preStopHook must delay Pod shut down sufficiently for both Drain latency and backend service drain timeout to complete, ensuring proper connection drainage and endpoint removal from the NEG before the Pod is shut down.

For optimal results, ensure your preStopHook execution time is less than or equal to the sum of the Backend Service Drain Timeout and Drain Latency.

This can be calculated using the formula:

preStopHook >= Backend Service Drain Timeout + Drain Latency

We recommend setting the Drain Latency to 1 minute. If 500 errors persist, estimate the total occurrence duration and add double that time to the latency. This ensures that your Pod has enough time to drain gracefully before being removed from the service. You can adjust this value if it's too long for your specific use case.

Alternatively, you can estimate the timing by examining the deletion timestamp from the Pod and the timestamp when the endpoint was removed from the NEG in the Cloud Audit Logs.

Termination Grace Period parameter

You must configure the terminationGracePeriod parameter to allow sufficient time for the preStopHook to finish and for the Pod to complete a graceful shutdown.

By default, when not explicitly set, the terminationGracePeriod is 30 seconds. You can calculate the optimal terminationGracePeriod using the formula:

terminationGracePeriod >= preStopHook Time + Pod shutdown time

To define terminationGracePeriod within the Pod's configuration as follows:

  spec:
    terminationGracePeriodSeconds: <terminationGracePeriod>
    containers:
    - name: my-app
      ...
    ...

NEG not found when creating an Internal Ingress resource

The following error might occur when you create an internal Ingress in GKE:

Error syncing: error running backend syncing routine: googleapi: Error 404: The resource 'projects/PROJECT_ID/zones/ZONE/networkEndpointGroups/NEG' was not found, notFound

This error occurs because Ingress for internal Application Load Balancers requires Network Endpoint Groups (NEGs) as backends.

In Shared VPC environments or clusters with Network Policies enabled, add the annotation cloud.google.com/neg: '{"ingress": true}' to the Service manifest.

504 Gateway Timeout: upstream request timeout

The following error might occur when you access a Service from an internal Ingress in GKE:

HTTP/1.1 504 Gateway Timeout
content-length: 24
content-type: text/plain

upsteam request timeout

This error occurs because traffic sent to internal Application Load Balancers are proxied by envoy proxies in the proxy-only subnet range.

To allow traffic from the proxy-only subnet range, create a firewall rule on the targetPort of the Service.

Error 400: Invalid value for field 'resource.target'

The following error might occur when you access a Service from an internal Ingress in GKE:

Error syncing:LB_NAME does not exist: googleapi: Error 400: Invalid value for field 'resource.target': 'https://www.googleapis.com/compute/v1/projects/PROJECT_NAME/regions/REGION_NAME/targetHttpProxies/LB_NAME. A reserved and active subnetwork is required in the same region and VPC as the forwarding rule.

To resolve this issue, create a proxy-only subnet.

Error during sync: error running load balancer syncing routine: loadbalancer does not exist

One of the following errors might occur when the GKE control plane upgrades or when you modify an Ingress object:

"Error during sync: error running load balancer syncing routine: loadbalancer
INGRESS_NAME does not exist: invalid ingress frontend configuration, please
check your usage of the 'kubernetes.io/ingress.allow-http' annotation."

Or:

Error during sync: error running load balancer syncing routine: loadbalancer LOAD_BALANCER_NAME does not exist:
googleapi: Error 400: Invalid value for field 'resource.IPAddress':'INGRESS_VIP'. Specified IP address is in-use and would result in a conflict., invalid

To resolve these issues, try the following steps:

  • Add the hosts field in the tls section of the Ingress manifest, then delete the Ingress. Wait five minutes for GKE to delete the unused Ingress resources. Then, recreate the Ingress. For more information, see The hosts field of an Ingress object.
  • Revert the changes you made to the Ingress. Then, add a certificate using an annotation or Kubernetes Secret.

Known issues

Cannot enable HTTPS Redirects with the V1 Ingress naming scheme

You cannot enable HTTPS redirects on GKE Ingress resources created on GKE versions 1.16.8-gke.12 and earlier. You must recreate the Ingress before you can enable HTTPS redirects, or an error event is created and the Ingress does not sync.

The error event message is similar to the following:

Error syncing: error running load balancer syncing routine: loadbalancer lb-name does not exist: ensureRedirectUrlMap() = error: cannot enable HTTPS Redirects with the V1 Ingress naming scheme. Please recreate your ingress to use the newest naming scheme.

Google Cloud Armor security policy fields removed from BackendConfig

There is a known issue where updating a BackendConfig resource using the v1beta1 API removes an active Google Cloud Armor security policy from its Service. This issue affects the following GKE versions:

  • 1.18.19-gke.1400 to 1.18.20-gke.5099
  • 1.19.10-gke.700 to 1.19.14-gke.299
  • 1.20.6-gke.700 to 1.20.9-gke.899

If you do not configure Google Cloud Armor on your Ingress resources via the BackendConfig then this issue does not affect your clusters.

For GKE clusters which do configure Google Cloud Armor through the BackendConfig, it is strongly recommended to only update BackendConfig resources using the v1 API. Applying a BackendConfig to your cluster using v1beta1 BackendConfig resources will remove your Google Cloud Armor security policy from the Service it is referencing.

To mitigate this issue, only make updates to your BackendConfig using the v1 BackendConfig API. The v1 BackendConfig supports all the same fields as v1beta1 and makes no breaking changes so the API field can be updated transparently. Replace the apiVersion field of any active BackendConfig manifests with cloud.google.com/v1 and do not use cloud.google.com/v1beta1. The following sample manifest describes a BackendConfig resource that uses the v1 API:

  apiVersion: cloud.google.com/v1
  kind: BackendConfig
  metadata:
    name: my-backend-config
  spec:
    securityPolicy:
      name: "ca-how-to-security-policy"

If you have CI/CD systems or tools which regularly update BackendConfig resources, ensure that you are using the cloud.google.com/v1 API group in those systems.

If your BackendConfig has already been updated with the v1beta1 API, your Google Cloud Armor security policy might have been removed. You can determine if this has happened by running the following command:

kubectl get backendconfigs -A -o json | jq -r '.items[] | select(.spec.securityPolicy == {}) | .metadata | "\(.namespace)/\(.name)"'

If the response returns output, then your cluster is impacted by the issue. The output of this command returns a list of BackendConfig resources (<namespace>/<name>) that are affected by the issue. If the output is empty, then your BackendConfig has not been updated using the v1beta1 API since the issue has been introduced. Any future updates to your BackendConfig should only use v1.

If your Google Cloud Armor security policy was removed, you can determine when it was removed using the following Logging query:

resource.type="gce_backend_service"
protoPayload.methodName="v1.compute.backendServices.setSecurityPolicy"
protoPayload.authenticationInfo.principalEmail:"container-engine-robot.iam.gserviceaccount.com"
protoPayload.response.status = "RUNNING"
NOT protoPayload.authorizationInfo.permission:"compute.securityPolicies.use"

If any of your clusters have been impacted, then this can be corrected by pushing an update to your BackendConfig resource that uses the v1 API.

Upgrade your GKE control plane to one of the following updated versions that patches this issue and allows v1beta1 BackendConfig resources to be used safely:

  • 1.18.20-gke.5100 and later
  • 1.19.14-gke.300 and later
  • 1.20.9-gke.900 and later

What's next