Release operator 2.6.0 (#1756)

* Initial move

* Continue removing model-in-image

* Model in image out of Java

* Remove model in image from introspector

* More corrections

* Continued removals

* Remove new schema elements

* Remove DomainSpec schema items added in 3.0

* Correct a couple more changes

* Domain status patching switching behavior

* Implement domain status switching behavior

* Bring over integration tests

* Bring more changes

* More integration tests

* Update domain1.yaml

* Update README

* Restore missing line

* Rebuild charts

* Support domain status detection in both directions

* Remove unnecessary variable

* Changes from develop

* Backport owls-80135

* Backport retry DB and RCU creation

* Rolling fixes

* Backport async HTTP client

* Backport 8.0.2 and fiber suspend debugging

* Port server status reader fixes

* Intermittent status fix

* Backport affinity merging

* WatchBuilderTest.tearDown() fix

* Node heating FAQ

* Sample updates

* OWLS-81419

* Introspector ssl issue

* Backport whitespace fix

* edit scaling doc as per OWLS-82641

* Backport cluster member sorting fix

* Backport new integration tests to the 2.6.0 branch. (#1728)

* Backport new integration test suite

* Adjust api version to v6

* Backport first half of check cluster visibility changes

* Include clustervie app

* Backport BigDecimal fix for ItPodsRestart

* Backport tail sooner enhancement

* Backport integration test cleanup improvements

* Backport upgrade instructions

* Backport OWLS-82320 namespace error

* Liveness probe tests

* Istio support

* CRD v7

* Correct for git names used as Docker image tags and cleanup repos where we are deleting the last tag

* Use GraalVM CE 19.3.2

* Reduce versions in CRDs

* Debugging around status endpoint change

* Accept null returned status with status endpoint

* Additional debugging

* Move initialization of using domain status endpoint

* add OLCNE 1.1 support (#1742)

* add OLCNE 1.1 support

* add K8s 1.18.0

* updated operator version to 2.6.0

* Backport OWLS-80884 fix to not stop pods no longer managed by this operator

* Corrections

* Review comments

* Fix JavaDoc

* Javadoc

* archive 2.5.0 docs

Signed-off-by: Mark Nelson <[email protected]>

* update readme current versions

Signed-off-by: Mark Nelson <[email protected]>

* archive 2.5.0 apidoc, chart, etc.

Signed-off-by: Mark Nelson <[email protected]>

Co-authored-by: Rosemary Marano <[email protected]>
Co-authored-by: Mark Nelson <[email protected]>

Author: 3 people

Dockerfile

@@ -4,37 +4,60 @@
 # HOW TO BUILD THIS IMAGE
 # -----------------------
 # Run:
-#      $ docker build -t weblogic-kubernetes-operator:latest .
+#      $ ./buildDockerImage.sh [-t <image-name>]
 #
-# Pull base image
-# From the Docker store
 # -------------------------
-FROM openjdk:11-oracle
-RUN yum -y install openssl && yum clean all
+FROM oraclelinux:7-slim
 
 # Maintainer
 # ----------
 MAINTAINER Ryan Eberhard <[email protected]>
 
-# make the operator run with a non-root user id (1000 is the `oracle` user)
-RUN groupadd -g 1000 oracle && \
-    useradd -d /operator -M -s /bin/bash -g 1000 -u 1000 oracle && \
-    mkdir /operator && \
-    mkdir /operator/lib && \
-    mkdir /logs && \
+RUN set -eux; \
+    yum -y install gzip tar openssl; \
+    rm -rf /var/cache/yum
+
+# Default to UTF-8 file.encoding
+ENV LANG en_US.UTF-8
+
+ENV JAVA_HOME /usr/local/graalvm-ce-java11
+ENV PATH /operator:$JAVA_HOME/bin:$PATH
+
+ENV JAVA_VERSION 11.0.7
+ENV JAVA_URL https://github.com/graalvm/graalvm-ce-builds/releases/download/vm-19.3.2/graalvm-ce-java11-linux-amd64-19.3.2.tar.gz
+
+# Install Java and make the operator run with a non-root user id (1000 is the `oracle` user)
+RUN set -eux; \
+    curl -fL -o /graalvm-ce-java11.tar.gz "$JAVA_URL"; \
+    mkdir -p "$JAVA_HOME"; \
+    tar --extract --file /graalvm-ce-java11.tar.gz --directory "$JAVA_HOME" --strip-components 1; \
+    rm /graalvm-ce-java11.tar.gz; \
+    mkdir /usr/java; \
+    ln -sfT "$JAVA_HOME" /usr/java/default; \
+    ln -sfT "$JAVA_HOME" /usr/java/latest; \
+    rm -Rf "$JAVA_HOME/include" "$JAVA_HOME/jmods" "$JAVA_HOME/languages" "$JAVA_HOME/tools" "$JAVA_HOME/lib/svm" "$JAVA_HOME/lib/installer" "$JAVA_HOME/lib/visualvm" "$JAVA_HOME/lib/truffle" "$JAVA_HOME/lib/polyglot"; \
+    rm -f "$JAVA_HOME/lib/src.zip" "$JAVA_HOME/lib/libjvmcicompiler.so" "$JAVA_HOME/bin/polyglot"; \
+    for bin in "$JAVA_HOME/bin/"*; do \
+        base="$(basename "$bin")"; \
+        [ ! -e "/usr/bin/$base" ]; \
+        alternatives --install "/usr/bin/$base" "$base" "$bin" 20000; \
+    done; \
+    java -Xshare:dump; \
+    groupadd -g 1000 oracle; \
+    useradd -d /operator -M -s /bin/bash -g 1000 -u 1000 oracle; \
+    mkdir -p /operator/lib; \
+    mkdir /logs; \
     chown -R 1000:1000 /operator /logs
-USER 1000
 
-ENV PATH=$PATH:/operator
+USER 1000
 
-ARG VERSION
 COPY src/scripts/* /operator/
-COPY operator/target/weblogic-kubernetes-operator-$VERSION.jar /operator/weblogic-kubernetes-operator.jar
+COPY operator/target/weblogic-kubernetes-operator.jar /operator/weblogic-kubernetes-operator.jar
 COPY operator/target/lib/*.jar /operator/lib/
 
 HEALTHCHECK --interval=1m --timeout=10s \
   CMD /operator/livenessProbe.sh
 
 WORKDIR /operator/
 
-CMD ["/operator/operator.sh"]
+CMD ["/operator/operator.sh"]

README.md

@@ -17,8 +17,8 @@ Oracle is finding ways for organizations using WebLogic Server to run important
 The fastest way to experience the operator is to follow the [Quick Start guide](https://oracle.github.io/weblogic-kubernetes-operator/quickstart/), or you can peruse our [documentation](https://oracle.github.io/weblogic-kubernetes-operator), read our [blogs](https://blogs.oracle.com/weblogicserver/updated-weblogic-kubernetes-support-with-operator-20), or try out the [samples](https://oracle.github.io/weblogic-kubernetes-operator/samples/).
 
 ***
-The [current release of the operator](https://github.com/oracle/weblogic-kubernetes-operator/releases) is 2.5.0.
-This release was published on February 26, 2020.
+The [current release of the operator](https://github.com/oracle/weblogic-kubernetes-operator/releases) is 2.6.0.
+This release was published on June 22, 2020.
 ***
 
 # Documentation

build-tools/src/main/resources/weblogic-kubernetes-operator/checkstyle/checkstyle-noframes-sorted.xsl renamed to build-tools/checkstyle/checkstyle-noframes-sorted.xsl

@@ -41,7 +41,7 @@
     </module>
 
     <module name="RegexpHeader">
-        <property name="headerFile" value="build-tools/src/main/resources/weblogic-kubernetes-operator/checkstyle/java.header"/>
+        <property name="header" value="^// Copyright \(c\) (\d\d\d\d, )+Oracle Corporation and\/or its affiliates\.$\n^// Licensed under the Universal Permissive License v 1\.0 as shown at https://oss\.oracle\.com/licenses/upl\.$"/>
         <property name="fileExtensions" value="java"/>
     </module>
 

build-tools/src/main/resources/weblogic-kubernetes-operator/checkstyle/customized_google_checks.xml renamed to build-tools/checkstyle/customized_google_checks.xml

@@ -0,0 +1,12 @@
+<?xml version="1.0"?>
+
+<!DOCTYPE suppressions PUBLIC
+    "-//Puppy Crawl//DTD Suppressions 1.1//EN"
+    "http://www.puppycrawl.com/dtds/suppressions_1_1.dtd">
+
+<suppressions>
+    <!-- There's nothing wrong with these - ignore them b/c they generate noise -->
+    <suppress checks="AbbreviationAsWordInName" files=".*"/>
+
+    <suppress checks="VariableDeclarationUsageDistanceCheck" files=".*"/>
+</suppressions>

build-tools/checkstyle/suppressions.xml

@@ -0,0 +1,88 @@
+#!/bin/bash
+# Copyright (c) 2020, Oracle Corporation and/or its affiliates.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl.
+
+usage() {
+cat << EOF
+
+Usage: buildDockerImage.sh [-t tag]
+Builds a Docker Image for the Oracle WebLogic Kubernetes Operator.
+
+Parameters:
+   -t: image name and tag in 'name:tag' format
+
+Copyright (c) 2020, Oracle Corporation and/or its affiliates.
+Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl.
+
+EOF
+exit 0
+}
+
+# WebLogic Kubernetes Operator Image Name
+name=""
+
+# Parameters
+while getopts "t:" optname; do
+  case ${optname} in
+    t )
+      name="$OPTARG"
+      ;;
+    \? )
+      usage
+      ;;
+  esac
+done
+
+IMAGE_NAME=${name:-oracle/weblogic-kubernetes-operator:2.6.0}
+SCRIPTPATH="$( cd "$(dirname "$0")" > /dev/null 2>&1 ; pwd -P )"
+
+# Proxy settings
+PROXY_SETTINGS=""
+if [ "${http_proxy}" != "" ]; then
+  PROXY_SETTINGS="$PROXY_SETTINGS --build-arg http_proxy=${http_proxy}"
+fi
+
+if [ "${https_proxy}" != "" ]; then
+  PROXY_SETTINGS="$PROXY_SETTINGS --build-arg https_proxy=${https_proxy}"
+fi
+
+if [ "${ftp_proxy}" != "" ]; then
+  PROXY_SETTINGS="$PROXY_SETTINGS --build-arg ftp_proxy=${ftp_proxy}"
+fi
+
+if [ "${no_proxy}" != "" ]; then
+  PROXY_SETTINGS="$PROXY_SETTINGS --build-arg no_proxy=${no_proxy}"
+fi
+
+if [ "$PROXY_SETTINGS" != "" ]; then
+  echo "Proxy settings were found and will be used during build."
+fi
+
+# ################## #
+# BUILDING THE IMAGE #
+# ################## #
+echo "Building image '$IMAGE_NAME' ..."
+
+# BUILD THE IMAGE (replace all environment variables)
+BUILD_START=$(date '+%s')
+docker build $PROXY_SETTINGS -t $IMAGE_NAME -f $SCRIPTPATH/Dockerfile $SCRIPTPATH || {
+  echo "There was an error building the image."
+  exit 1
+}
+BUILD_END=$(date '+%s')
+BUILD_ELAPSED=`expr $BUILD_END - $BUILD_START`
+
+echo ""
+
+if [ $? -eq 0 ]; then
+cat << EOF
+  WebLogic Kubernetes Operator Docker Image is ready:
+
+    --> $IMAGE_NAME
+
+  Build completed in $BUILD_ELAPSED seconds.
+
+EOF
+else
+  echo "WebLogic Kubernetes Operator Docker Image was NOT successfully created. Check the output and correct any reported problems with the docker build operation."
+fi

build-tools/src/main/resources/weblogic-kubernetes-operator/pmd/weblogicx-ruleset.xml renamed to build-tools/pmd/weblogicx-ruleset.xml

@@ -8,7 +8,7 @@
     <parent>
         <artifactId>operator-parent</artifactId>
         <groupId>oracle.kubernetes</groupId>
-        <version>2.5.0</version>
+        <version>2.6.0</version>
     </parent>
 
     <artifactId>buildtime-reports</artifactId>
@@ -63,5 +63,11 @@
             <artifactId>operator-integration-tests</artifactId>
             <version>${project.version}</version>
         </dependency>
+        <dependency>
+            <groupId>${project.groupId}</groupId>
+            <artifactId>new-integration-tests</artifactId>
+            <version>${project.version}</version>
+        </dependency>
     </dependencies>
+
 </project>

build-tools/src/main/resources/weblogic-kubernetes-operator/pmd/weblogicx-test-ruleset.xml renamed to build-tools/pmd/weblogicx-test-ruleset.xml

@@ -23,8 +23,8 @@ using the operator to deploy and run a WebLogic domain container-packaged web ap
 ***
 #### Current production release
 
-The [current production release of the operator](https://github.com/oracle/weblogic-kubernetes-operator/releases) is 2.5.0.
-This release was published on February 26, 2020. See the operator prerequisites and supported environments [here]({{< relref "/userguide/introduction/introduction#operator-prerequisites" >}}).
+The [current production release of the operator](https://github.com/oracle/weblogic-kubernetes-operator/releases) is 2.6.0.
+This release was published on June 22, 2020. See the operator prerequisites and supported environments [here]({{< relref "/userguide/introduction/introduction#operator-prerequisites" >}}).
 
 #### Preview of next major release
 
@@ -33,7 +33,7 @@ This release candidate is suitable for use by early adopters who wish to test 3.
 This release candidate was published on May 8, 2020.  There may be additional release candidates before the final 3.0.0 release.
 
 This release candidate introduces _non-backward compatible_ changes.  This release candidate cannot be run in the same
-cluster as another release of the operator.  You can upgrade from 2.5.0 to 3.0.0-rc1 without needing to restart or recreate
+cluster as another release of the operator.  You can upgrade from 2.6.0 to 3.0.0-rc1 without needing to restart or recreate
 any existing domains. However, please note that we _do_ plan to support running the final 3.0.0
 release in the same cluster with at least one 2.x release of the operator to allow for staged migration.
 

build-tools/pom.xml

@@ -91,7 +91,7 @@ Here is an example of part of a domain custom resource file with the `imagePullS
 specified:
 
 ```
-apiVersion: "weblogic.oracle/v2"
+apiVersion: "weblogic.oracle/v7"
 kind: Domain
 metadata:
   name: domain1
@@ -101,7 +101,7 @@ metadata:
     weblogic.domainUID: domain1
 spec:
   domainHome: /u01/oracle/user_projects/domains/domain1
-  domainHomeSourceType: Image
+  domainHomeInImage: true
   image: "some.registry.com/owner/domain1:1.0"
   imagePullPolicy: "IfNotPresent"
   imagePullSecrets:

build-tools/src/main/resources/weblogic-kubernetes-operator/spotbugs/exclude-common.xml renamed to build-tools/spotbugs/exclude-common.xml

@@ -39,7 +39,7 @@ elkIntegrationEnabled: false
 externalDebugHttpPort: 30999
 externalRestEnabled: false
 externalRestHttpsPort: 31001
-image: oracle/weblogic-kubernetes-operator:2.5.0
+image: oracle/weblogic-kubernetes-operator:2.6.0
 imagePullPolicy: IfNotPresent
 internalDebugHttpPort: 30999
 istioEnabled: false

build-tools/src/main/resources/weblogic-kubernetes-operator/checkstyle/java.header

@@ -0,0 +1,55 @@
+---
+title: "Node heating problem"
+date: 2020-06-03T08:08:19-04:00
+draft: false
+weight: 22
+---
+
+The WebLogic Server Kubernetes Operator creates a Pod for each WebLogic Server instance that is started. The [Kubernetes Scheduler](https://kubernetes.io/docs/concepts/scheduling-eviction/kube-scheduler/) then selects a Node for each Pod. Because the default scheduling algorithm gives substantial weight to selecting a Node where the necessary Docker images have already been pulled, this often results in Kubernetes running many of the Pods for WebLogic Server instances on the same Node while other Nodes are not fairly utilized.
+
+This is commonly known as the "Node heating problem." One solution is to ensure that all necessary Docker images are available on worker Nodes as part of node provisioning. When the necessary Docker images are available on each worker Node, the Kubernetes Scheduler will instead select a Node based on other factors such as available CPU and memory or a simple round-robin.
+
+The operator team recommends a different solution that is based on [inter-pod affinity and anti-affinity](https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#inter-pod-affinity-and-anti-affinity). This solution has the advantage of both resolving the Node heating problem and of explicitly directing the Kubernetes Scheduler to spread the Pods for WebLogic Server instances from a given cluster or domain more widely across the available Nodes. Inter-pod affinity and anti-affinity are features of the Kubernetes Scheduler that allow the scheduler to choose a Node for a new Pod based on details of the Pods that are already running. For WebLogic Server use cases, the intent will often be for anti-affinity with the Pods for other WebLogic Server instances so that server instances spread over the available Nodes.
+
+To use these features, edit the Domain Custom Resource to add content to the `serverPod` element, in this case at the scope of a cluster, as shown in the following example:
+
+```
+clusters:
+- clusterName: cluster-1
+  serverStartState: "RUNNING"
+  serverPod:
+    affinity:
+      podAntiAffinity:
+        preferredDuringSchedulingIgnoredDuringExecution:
+          - weight: 100
+            podAffinityTerm:
+              labelSelector:
+                matchExpressions:
+                  - key: "weblogic.clusterName"
+                    operator: In
+                    values:
+                    - $(CLUSTER_NAME)
+              topologyKey: "kubernetes.io/hostname"
+```
+
+Because the `serverPod` element here is scoped to a cluster, the content of the `affinity` element will be added to the Pod generated for each WebLogic Server instance that is a member of this WebLogic cluster. This inter-pod anti-affinity statement expresses a preference that the scheduler select a Node for the new Pod avoiding, as much as possible, Nodes that already have Pods with the label "weblogic.clusterName" and the name of this cluster. Note that the `weight` is set to `100`, which is the maximum weight, so that this term will outweigh any possible preference for a Node based on availability of Docker images.
+
+It is possible to express many other scheduling preferences or constraints. The following example similarly expresses an anti-affinity, but changes the test to have all WebLogic Server instances in the domain prefer to run on Nodes where there is not already a Pod for a running instance:
+
+```
+serverPod:
+  affinity:
+    podAntiAffinity:
+      preferredDuringSchedulingIgnoredDuringExecution:
+      - weight: 100
+        podAffinityTerm:
+          labelSelector:
+            matchExpressions:
+            - key: "weblogic.domainUID"
+              operator: In
+              values:
+              - $(DOMAIN_UID)
+          topologyKey: "kubernetes.io/hostname"
+```
+
+Details about how the operator generates Pods for WebLogic Server instances, including details about labels and variable substitution, are available [here]({{< relref "/userguide/managing-domains/domain-resource#pod-generation" >}}).

build-tools/src/main/resources/weblogic-kubernetes-operator/checkstyle/suppressions.xml

@@ -19,7 +19,7 @@ and accept the license agreement for the [WebLogic Server image](https://hub.doc
 1. Pull the operator image:
 
     ```bash
-    $ docker pull oracle/weblogic-kubernetes-operator:2.5.0
+    $ docker pull oracle/weblogic-kubernetes-operator:2.6.0
     ```
 
     {{% notice note %}} If you are here because you are following the Model In Image sample,

buildDockerImage.sh

@@ -71,7 +71,7 @@ $ helm install traefik-operator stable/traefik \
     ```bash
     $ helm install sample-weblogic-operator kubernetes/charts/weblogic-operator \
       --namespace sample-weblogic-operator-ns \
-      --set image=oracle/weblogic-kubernetes-operator:2.5.0 \
+      --set image=oracle/weblogic-kubernetes-operator:2.6.0 \
       --set serviceAccount=sample-weblogic-operator-sa \
       --set "domainNamespaces={}" \
       --wait

buildtime-reports/pom.xml

@@ -8,8 +8,9 @@ draft: false
 
 | Date | Version | Introduces backward incompatibilities | Change |
 | --- | --- | --- | --- |
-| May 8, 2020 | v3.0.0-rc1 | yes | Adds Model in Image feature. Support for Kubernetes 1.16. Removal of support for Kubernetes 1.13 and earlier. Removal of support for Helm 2.x. This release candidate cannot be run in the same cluster as another release of the operator.  You can upgrade from 2.5.0 to 3.0.0-rc1 without needing to restart or recreate any existing domains. However, please note that we do plan to support running the final 3.0.0 release in the same cluster with at least one 2.x release of the operator to allow for staged migration. |
-| February 26, 2020 | v2.5.0 | no | Support for Helm 3.x and OpenShift 4.3.  Operator can be installed in a namespace-dedicated mode where operator requires no cluster-level Kubernetes privileges. This version is not supported on Kubernetes 1.16+, check the [prerequisites]({{< relref "/userguide/introduction/introduction#operator-prerequisites" >}}).
+| June 22, 2020 | v2.6.0 | no | Kubernetes 1.16, 1.17, and 1.18 support. Removal of support for Kubernetes 1.13 and earlier. This release can be run in the same cluster with operators of either 2.5.0 and below, or with 3.x providing an upgrade path. Certified support of Oracle Linux Cloud Native Environment (OLCNE) 1.1 with Kubernetes 1.17.0.
+| May 8, 2020 | v3.0.0-rc1 | yes | Adds Model in Image feature. Support for Kubernetes 1.16. Removal of support for Kubernetes 1.13 and earlier. Removal of support for Helm 2.x. This release candidate cannot be run in the same cluster as another release of the operator.  You can upgrade from the current release of the operator to 3.0.0-rc1 without needing to restart or recreate any existing domains. However, please note that we _do_ plan to support running the final 3.0.0 release in the same cluster with at least one 2.x release of the operator to allow for staged migration. |
+| February 26, 2020 | v2.5.0 | no | Support for Helm 3.x and OpenShift 4.3.  Operator can be installed in a namespace-dedicated mode where operator requires no cluster-level Kubernetes privileges. This version is not supported on Kubernetes 1.16+; check the [prerequisites]({{< relref "/userguide/introduction/introduction#operator-prerequisites" >}}).
 | November 15, 2019 | v2.4.0 | no | Includes fixes for a variety of issues related to FMW infrastructure domains and pod variable substitution.  Operator now uses WebLogic Deploy Tooling 1.6.0 and the latest version of the Kubernetes Java Client.
 | August 27, 2019 | v2.3.0 | no  | Added support for Coherence cluster rolling, pod templating and additional pod content, and experimental support for running under an Istio service mesh.
 | June 20, 2019 | v2.2.1 | no  | The operator now supports Kubernetes 1.14.0+.  This release is primarily a bug fix release and resolves the following issues:<br><ul><li>Servers in domains, where the domain home is on a persistent volume, would sometimes fail to start. These failures would be during the introspection phase following a full domain shutdown.  Now, the introspection script better handles the relevant error conditions.</li><li>The domain resource provides an option to [pre-create Kubernetes Services](https://github.com/oracle/weblogic-kubernetes-operator/blob/master/docs/domains/Domain.md#server-service) for WebLogic Servers that are not yet running so that the DNS addresses of these services are resolvable.  These services are now created as non-headless so that they have an IP address.</li></ul>