Operator 4.0.4

• Resolved an issue related to the logHomeLayout field during schema conversion from "v8" to "v9" Domains.
  • Operator 4.0 introduced the logHomeLayout field to control the organization of WebLogic Server instance log files.
  • The default value for logHomeLayout is ByServers; however, the operator 3.4 behavior was consistent with the value of Flat.
  • Therefore, schema conversion from "v8" to "v9" Domains now explicitly selects a logHomeLayout value of Flat.
• Resolved an issue during uninstall where the validating webhook was leaving behind a ValidatingWebhookConfiguration resource. This resource is now deleted.

Operator 4.0.3

• The Helm chart now excludes the operator's webhook from running in an Istio mesh.
• Communication from the operator to WebLogic Server instances or to the WebLogic Monitoring Exporter sidecar use the service name and namespace rather than IP addresses to be more compatible with Istio strict mutual TLS mode.
• Resolved an issue where the Completed status condition would be set to true too early during the scale down of a cluster to 0 replicas.
• The Model in Image sample now uses "v9" style auxiliary images.
• The coordination between the periodic listing of Domain and Cluster resources and the processing of watch notifications has been improved to increase operator efficiency.

Operator 4.0.2

• The Helm chart now supports configuring tolerations for the operator's deployment.
• Domain resource schema now supports .spec.maxClusterUnavailable, which is the default value for the maximum number of cluster members that can be temporarily unavailable, such as during a roll.
• Resolved an issue that would prevent Model in Image domains from working correctly when Secured Production Mode is enabled.
• Pods created for WebLogic Server instances that are members of a WebLogic cluster will now include the label weblogic.clusterObservedGeneration specifying the metadata.generation of the Cluster resource. This is similar to the already existing weblogic.domainObservedGeneration label that specifies the observed generation of the Domain resource.
• Schema conversion from "v8" to "v9" Domains now uses shorter name prefixes when converting "v8" style auxiliary images.
• The Helm chart specifies a more compatible securityContext for the operator's deployment when the kubernetesPlatform value is "OpenShift".
• The Node Manager instance running within the Pod created for each WebLogic Server instance now supports additional configuration for how often to restart WebLogic before the Pod's liveness probe fails and the Pod is recreated.
• Samples and utility scripts now support the environment variables KUBERNETES_CLI, which defaults to kubectl, and WLSIMG_BUILDER, which defaults to docker, to make it easier to use these samples and scripts on platforms that use binaries with different names.
• Resolved an issue where a "v8" Domain that used auxiliary images and also configured logHome could be converted to a "v9" Domain that was invalid.
• Resolved an issue where Domains that had a .spec.domainUid that was different from the value of .metadata.name would not start.

Operator 4.0.1

• Resolved an issue where introspection would fail because the function wlsVersionEarlierThan was missing.

Operator 4.0.0

Major Themes

• Auxiliary Image simplification:

  • The usage of Model in Image auxiliary images has been substantially simplified.
  • In most cases, customers now need only specify the name of the auxiliary image.

• New Cluster resource:

  • The operator now provides a new custom resource, Cluster, which configures a specific WebLogic cluster and provides status information for that cluster.
  • This resource can be used with Kubernetes Horizontal Pod Autoscaling (HPA) or similar technologies to scale WebLogic clusters.

• Status and Events updates:

  • The Conditions and Events generated about Domain and Cluster resources have been significantly updated.
  • The operator more clearly communicates when the domain or individual clusters have reached the intended state or significant availability for your application workloads.
  • Failures are distinguished between those that require customer intervention and those that are potentially temporary and will be retried.

• Retry updates:

  • The operator handling of retries after a failure have been significantly updated.
  • Customers can now tune retry intervals and limits, or disable retries, on a per domain basis.
  • The time of the first failure, last retry, and the expected time of the next retry can now be easily inferred from Domain resource .status.

• Istio compatibility:

  • Domain configurations are always compatible with Istio and other service mesh products.
  • Customers are no longer required to enable this support.

Upgrade Notes

• This release introduces a new API version for the Domain resource, weblogic.oracle/v9.

  • The previous API version, weblogic.oracle/v8 introduced with Operator 3.0.0 is deprecated, but is still supported.

• This release also introduces a new custom resource, the weblogic.oracle/v1 Cluster resource.

  • This Cluster resource configures a specific WebLogic cluster and allows customers to scale that cluster using the Kubernetes Horizontal Pod Autoscaling (HPA) or similar technologies.
  • The Cluster resource .spec schema is similar to and replaces the former weblogic.oracle/v8 Domain resource .spec.clusters[*] content.
  • The Cluster resource .status schema mirrors the content of the Domain resource .status.clusters[*] content.

• This release simplifies upgrading the operator.

  • The operator added a schema conversion webhook.
    • This webhook automates the upgrade of weblogic.oracle/v8 Domain resources to weblogic.oracle/v9.
    • Customers do not need to take any action to begin using the latest Domain schema; however, the team recommends starting with "v9" Domain resources for any new projects.
  • Customers who install a single WebLogic Kubernetes Operator per Kubernetes cluster (most common use case) can upgrade directly from any 3.x operator release to 4.0.0. The Helm chart for 4.0.0 will automatically install the schema conversion webhook.
  • For those customers who install more than one WebLogic Kubernetes Operator in a single Kubernetes cluster:
    • You must upgrade every operator to at least version 3.4.1 before upgrading any operator to 4.0.0.
    • As the 4.0.0 Helm chart now also installs a singleton schema conversion webhook that is shared by all 4.0.0 operators in the cluster, you will want to use the webhookOnly Helm chart option to install this webhook in its own namespace prior to installing any of the 4.0.0 operators, and also use the preserveWebhook Helm chart option with each operator to prevent an operator uninstall from uninstalling the shared webhook.
  • The operator provides a utility that can be used to convert existing "v8" Domain YAML files to "v9".
  • Several [Helm chart default values have been changed]({{< ref "#changes-to-operator-helm-chart-configuration-values" >}}). Customers who upgrade their 3.x installations using the --reuse-values option during the Helm upgrade will continue to use the values from their original installation.

Changes to domain schema

• Model in Image auxiliary image related.

  • Added .spec.configuration.model.auxiliaryImages for simplified configuration of auxiliary images.
  • Added several additional advanced settings related to auxiliary images including .spec.configuration.model.auxiliaryImageVolumeMountPath, .spec.configuration.model.auxiliaryImageVolumeMedium, and .spec.configuration.model.auxiliaryImageVolumeSizeLimit.
  • Removed .spec.serverPod.auxiliaryImages and .spec.auxiliaryImageVolumes as part of the simplification effort.

• Cluster changes.

  • Moved content under .spec.clusters[*] to the new Cluster resource .spec.
  • Modified .spec.clusters[*] to be a list of Cluster resource names.
  • Removed allowReplicasBelowMinDynClusterSize field from .spec. and .spec.clusters[*] as cluster minimums must now be configured through autoscaling.

• Retry and failure tuning related.

  • Added .spec.failureRetryIntervalSeconds and .spec.failureRetryLimitMinutes for configuration of failure retry timing.
  • Added .spec.serverPod.maxReadyWaitTimeSeconds for configuring maximum time before considering servers as stalled during startup.

• Condition status updates.

  • Added or updated Conditions to clearly reflect the current state of the domain:
    • The Completed condition is set to True when no failures are detected, plus all expected pods are ready, at their target image(s), restartVersion, and introspectVersion.
    • The Available condition is set to True when a sufficient number of pods are ready. This condition can be True even when Completed is False, a Failed condition is reported, or a cluster has up to cluster.spec.maxUnavailable pods that are not ready.
    • The Failed condition is set to True and its .message field is updated when there is a failure.
  • Added .status.conditions[*].severity to describe the severity of conditions that represent failures or warnings.
  • Added fields .status.initialFailureTime and .status.lastFailureTime to enable measuring how long a Failure condition has been reported.
  • Removed condition ServersReady. Instead use Available or Completed.
  • Removed condition Progressing. Instead:
    • Check Completed is True to see when a Domain is fully up-to-date and running.
    • Check whether Available or Completed exist to verify that the operator has noticed the Domain resource.
    • If Failed is True, then check its .message and .severity to see if retries have stopped.

• General status updates.

  • Added .status.observedGeneration for comparison to .metadata.generation; if the two values match, then the status correctly reflects the status of the latest updates to the Domain resource.
  • Renamed .status.lastIntrospectJobProcessedUid to .status.failedIntrospectionUid.
  • Added .status.serverStatus[*].podReady and .status.serverStatus[*].podPhase with additional information about the pod associated with each WebLogic Server instance.

• Miscellaneous changes.

  • Added .spec.logHomeLayout for configuring the directory layout for WebLogic Server logs.
  • Added .spec.serverPod.shutdown.waitForAllSessions to enhance the configuration of server shutdown options.
  • Removed .spec.configuration.istio as this section is no longer required.
  • Removed previously deprecated fields .spec.domainHomeInImage, .spec.configOverrides, and .spec.configOverrideSecrets.
    These have been replaced with .spec.domainHomeSourceType, .spec.configuration.overridesConfigMap, and .spec.configuration.secrets.
  • Removed serverStartState field from .spec, .spec.adminServer, .spec.managedServers[*], and spec.clusters[*] as this field is no longer needed.
  • Default change for .spec.adminServer.adminChannelPortForwardingEnabled to true. It previously defaulted to false.

Changes to operator Helm chart configuration values

• Changed defaults for configuration values.

  • The default for domainNamespaceSelectionStrategy is now LabelSelector. It was previously List.
  • The default for domainNamespaceLabelSelector is now weblogic-operator=enabled. It was previously unspecified.
  • The default for enableClusterRoleBinding is now true; however, this value is ignored if domainNamespaceSelectionStrategy is Dedicated. It was previously false.

• Removed configuration value.

  • The previously deprecated value dedicated has been removed. Set the domainNamespaceSelectionStrategy to Dedicated instead.

• New configuration values.

  • Added value createLogStashConfigMap, default true, that configures how an optional Logstash container for the operator is configured.
  • Added value webhookOnly, default false, that specifies if this Helm release will install only the schema conversion webhook.
  • Added value operatorOnly, default false, that specifies if this Helm release will install only the operator.
  • Added value `preserveWeb...

Operator 3.4.4

• Support added to specify resource requests and limits for Monitoring Exporter sidecar containers.
• Resolved an issue that prevented the generation of the RollCompleted event.
• Updated scaling and lifecycle scripts to work correctly when used in a cluster with the upcoming WebLogic Kubernetes Operator 4.0 release.
• Updated several dependencies, including SnakeYAML and the Oracle Linux base for the container image.

Operator 3.4.3

• Resolved an issue where introspection would fail due to the node manager for non-English locales (#3328).
• Resolved an issue related to configuration override files where servers may fail to start because another server was updating files on a shared volume (#3359).

Operator 3.4.2

• Updated several dependencies, including the Oracle Linux base for the container image.
• Updated the default WebLogic Monitoring Exporter injection to version 2.0.7.

Operator 3.4.1

• Better support for retrying introspector DeadlineExceeded failures (#3109).
• For Model in Image, expand the WDT custom folder from the archive in the domain directory before calling update or create domain (#3110).

Operator 3.4.0

• Certified support for Kubernetes 1.22.5+ and 1.23.4+.
• Support for injecting Fluentd sidecar containers for WebLogic Server pods and the introspector.
• Resolved an issue where operator Services caused warnings in the Kiali console because of missing protocol designations (#2957)
• Updated multiple dependencies, including updating the default WebLogic Monitoring Exporter injection to version 2.0.5.
• Resolved issue #2794 related to restarting evicted pods (#2979).