OpenShift Serverless Release Notes

OpenShift Serverless now uses Knative Serving 0.19.0.

OpenShift Serverless uses Knative Eventing 0.19.2.

OpenShift Serverless now uses Kourier 0.19.0.

OpenShift Serverless now uses Knative kn CLI 0.19.1.

OpenShift Serverless now uses Knative Kafka 0.19.1.

A DomainMapping custom resource (CR) has been added to OpenShift Serverless to enable users to map a custom domain name to a Knative Service. See the Knative documentation on Creating a mapping between a custom domain name and a Knative Service.

In Knative Serving 0.19.0, v1alpha1 and v1beta1 versions of the Service, Route, Configuration, and Revision resources have been removed. The OpenShift Serverless Operator automatically upgrades older resources to v1, so no user action is required.

OpenShift Serverless now uses Knative Serving 0.18.2.

OpenShift Serverless uses Knative Eventing 0.18.6.

OpenShift Serverless now uses Kourier 0.18.0.

OpenShift Serverless now uses Knative kn CLI 0.18.4.

OpenShift Serverless now uses Knative Kafka 0.18.0.

In previous versions, if you used a ping source with OpenShift Serverless, after you uninstalled and deleted all other Knative Eventing components, the pingsource-jobrunner deployment was not deleted. This issue is now fixed, and the pingsource-jobrunner deployment has been renamed to pingsource-mt-adapter.

In previous versions, deleting a sink before you delete the SinkBinding resource connected to it caused the resource deletion to hang. This issue is now fixed.

Using the eventing.knative.dev/scope: namespace annotation for the KafkaChannel objects is not supported.

Knative Eventing on OpenShift Serverless is now Generally Available (GA).

Apache Kafka features such as Kafka channel and Kafka event source are now available as a Technology Preview on OpenShift Serverless. Kafka integration is delivered through the OpenShift Serverless Operator and does not require a separate community Operator installation. For more information, see the documentation on Using Apache Kafka with OpenShift Serverless.

OpenShift Serverless Functions is now delivered as a Developer Preview through the standard Knative kn CLI installation. This feature is not yet supported by Red Hat for production deployments, but can be used for development and testing. For more information about using OpenShift Serverless Functions through the kn func CLI plug-in, see the OpenShift Serverless Functions Developer Preview documentation.

OpenShift Serverless now uses Knative Serving 0.17.3.

OpenShift Serverless uses Knative Eventing 0.17.2.

OpenShift Serverless now uses Kourier 0.17.0.

OpenShift Serverless now uses Knative kn CLI 0.17.3.

OpenShift Serverless now uses Knative Kafka 0.17.1.

When the horizontal pod autoscaler (HPA) scales up the broker-ingress pod, the imc-dispatcher pod sometimes fails to forward replies. This is because the new broker-ingress pods are Ready before accepting connections, because they lack a readiness probe. If you are using HPA autoscaling and do not want to scale the broker-ingress pod manually, you must configure retries in the Broker.Spec.Delivery.

Using the eventing.knative.dev/scope: namespace annotation with Kafka channels is not supported.

OpenShift Serverless now uses Knative Operator 0.16.0.

OpenShift Serverless now uses Knative Serving 0.16.0.

OpenShift Serverless uses Knative Eventing 0.16.0.

OpenShift Serverless now uses Kourier 0.16.0.

OpenShift Serverless now uses Knative kn CLI 0.16.1.

The annotation knative-eventing-injection=enabled that was previously used to label namespaces for broker creation is now deprecated. The new annotation is eventing.knative.dev/injection=enabled.

Multi-container support is now available on Knative as a Technology Preview feature. You can enable multi-container support in the config-features config map. For more information, see the Knative documentation.

In previous releases, Knative Serving had a fixed, minimum CPU request of 25m for queue-proxy. If your cluster had any value set that conflicted with this, for example, if you had set a minimum CPU request for defaultRequest of more than 25m, the Knative Service failed to deploy. This issue is fixed in 1.10.0.