How to add Kubernetes Gateway API features to an ingress requirer charm¶

Charms that implement only the ingress relation can leverage the Kubernetes Gateway API with the help of the ingress-configurator charm.

Deploy the charms¶

Deploy the ingress requirer charm:

juju deploy <ingress-requirer-charm>

If your charm has a separate action or configuration step to start the workload, run it now and wait until the unit is active.

This guide assumes the ingress requirer charm already works correctly.

Deploy the gateway-api-integrator charm:

juju deploy gateway-api-integrator --channel=latest/edge --trust

Deploy the ingress-configurator charm:

juju deploy ingress-configurator --channel=edge --trust

The --trust flag is required to allow ingress-configurator to manage Kubernetes resources in the cluster.

Integrate the ingress-configurator charm with both the gateway-api-integrator charm and the ingress requirer:

juju integrate ingress-configurator:gateway-route gateway-api-integrator
juju integrate ingress-configurator:ingress <ingress-requirer-charm>

Configure a hostname:

HOSTNAME=<hostname>
juju config ingress-configurator hostname=${HOSTNAME}

To expose the workload on additional hostnames, configure additional-hostnames:

juju config ingress-configurator additional-hostnames=<hostname1>,<hostname2>

To restrict routing to specific URL paths, configure paths:

juju config ingress-configurator paths=<path1>,<path2>

Use the get-proxied-endpoints action on ingress-configurator:

juju run ingress-configurator/leader get-proxied-endpoints

You should see https://${HOSTNAME}/ in the endpoints field.

If additional-hostnames is configured, those hostnames should also appear in the endpoints field.

To send a request directly to the gateway, first retrieve the gateway’s external IP address:

GATEWAY_IP=$(kubectl get gateway -A -o jsonpath='{.items[0].status.addresses[0].value}')

Then send a request using the configured hostname:

curl -i "http://${GATEWAY_IP}/" -H "Host: ${HOSTNAME}"

Adjust the path to match an endpoint exposed by your backend workload.

You should see a response from the backend workload, confirming that the gateway is correctly routing traffic to it.