build: optimizations for incremental builds and faster CI (#13033)

Co-authored-by: Andrew Sikowitz <andrew.sikowitz@acryl.io>

.github/workflows/docker-unified.yml

@@ -447,7 +447,7 @@ jobs:
       - name: Pre-build artifacts for docker image
         run: |
           ./gradlew :datahub-frontend:dist -x test -x yarnTest -x yarnLint --parallel
-          mv ./datahub-frontend/build/distributions/datahub-frontend-*.zip datahub-frontend.zip
+          mv ./datahub-frontend/build/stage/main .
       - name: Build and push
         uses: ./.github/actions/docker-custom-build-and-push
         with:

build.gradle

@@ -91,7 +91,7 @@ plugins {
   // TODO id "org.gradlex.java-ecosystem-capabilities" version "1.0"
 }
 
-apply from: "gradle/docker/docker.gradle"
+apply from: "gradle/docker/docker-utils.gradle"
 
 project.ext.spec = [
     'product' : [
@@ -112,12 +112,12 @@ project.ext.spec = [
 
 project.ext.externalDependency = [
     'akkaHttp': "com.typesafe.akka:akka-http-core_$playScalaVersion:10.2.10", // max version due to licensing
+    'akkaParsing': "com.typesafe.akka:akka-parsing_$playScalaVersion:10.2.10", // akka-parsing is part of akka-http, so use akka http version
     'akkaActor': "com.typesafe.akka:akka-actor_$playScalaVersion:$akkaVersion",
     'akkaStream': "com.typesafe.akka:akka-stream_$playScalaVersion:$akkaVersion",
     'akkaActorTyped': "com.typesafe.akka:akka-actor-typed_$playScalaVersion:$akkaVersion",
     'akkaSlf4j': "com.typesafe.akka:akka-slf4j_$playScalaVersion:$akkaVersion",
     'akkaJackson': "com.typesafe.akka:akka-serialization-jackson_$playScalaVersion:$akkaVersion",
-    'akkaParsing': "com.typesafe.akka:akka-parsing_$playScalaVersion:$akkaVersion",
     'akkaProtobuf': "com.typesafe.akka:akka-protobuf-v3_$playScalaVersion:$akkaVersion",
     'antlr4Runtime': 'org.antlr:antlr4-runtime:4.9.3',
     'antlr4': 'org.antlr:antlr4:4.9.3',

datahub-frontend/build.gradle

@@ -1,12 +1,12 @@
 plugins {
   id 'scala'
-  id 'com.palantir.docker'
   id 'org.gradle.playframework'
 }
 
 apply from: '../gradle/versioning/versioning.gradle'
 apply from: './play.gradle'
 apply from: '../gradle/coverage/java-coverage.gradle'
+apply from: '../gradle/docker/docker.gradle'
 
 ext {
   docker_repo = 'datahub-frontend-react'
@@ -34,7 +34,7 @@ model {
 }
 
 task myTar(type: Tar) {
-  compression = Compression.GZIP
+  compression = Compression.NONE
 
   from("${buildDir}/stage")
 
@@ -66,10 +66,19 @@ distributions {
   }
 }
 
+task unversionZip(type: Copy, dependsOn: [':datahub-web-react:distZip', distZip]) {
+
+  from ("${buildDir}/distributions")
+  include "datahub-frontend-${version}.zip"
+  into "${buildDir}/distributions"
+  rename "datahub-frontend-${version}.zip", "datahub-frontend.zip"
+}
+
 docker {
+  dependsOn(stageMainDist)
   name "${docker_registry}/${docker_repo}:v${version}"
-  version "v${version}"
   dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
+  files "${buildDir}/stage"
   files fileTree(rootProject.projectDir) {
     include '.dockerignore'
     include 'docker/monitoring/*'
@@ -77,12 +86,7 @@ docker {
   }.exclude {
     i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
   }
-  tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-  // platform('linux/arm64', 'linux/amd64')
-  buildx(true)
-  load(true)
-  push(false)
+  additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
   // Add build args if they are defined (needed for some CI or enterprise environments)
   def dockerBuildArgs = [:]
@@ -101,20 +105,6 @@ docker {
   }
 }
 
-task unversionZip(type: Copy, dependsOn: [':datahub-web-react:distZip', dist]) {
-  from ("${buildDir}/distributions")
-  include "datahub-frontend-${version}.zip"
-  into "${buildDir}/docker/"
-  rename "datahub-frontend-${version}.zip", "datahub-frontend.zip"
-}
-tasks.getByPath(":datahub-frontend:docker").dependsOn(unversionZip)
-
-task cleanLocalDockerImages {
-  doLast {
-    rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-  }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)
 
 // gradle 8 fixes
 tasks.getByName('createDatahub-frontendTarDist').dependsOn 'stageMainDist'

datahub-graphql-core/build.gradle

@@ -1,6 +1,6 @@
 plugins {
     id 'java'
-    id "io.github.kobylynskyi.graphql.codegen" version "4.1.1"
+    id "io.github.kobylynskyi.graphql.codegen" version "5.10.0"
 }
 
 apply from: '../gradle/coverage/java-coverage.gradle'
@@ -41,6 +41,7 @@ graphqlCodegen {
     generateApis = true
     generateParameterizedFieldsResolvers = false
     modelValidationAnnotation = "@javax.annotation.Nonnull"
+    addGeneratedAnnotation = false  // Skips timestamps in generated files which forces re-compile
     customTypesMapping = [
         Long: "Long",
         Float: "Float"

datahub-upgrade/build.gradle

@@ -1,11 +1,11 @@
 plugins {
   id 'org.springframework.boot'
   id 'java'
-  id 'com.palantir.docker'
 }
 
 apply from: "../gradle/versioning/versioning.gradle"
 apply from: "../gradle/coverage/java-coverage.gradle"
+apply from: "../gradle/docker/docker.gradle"
 
 ext {
   docker_registry = rootProject.ext.docker_registry == 'linkedin' ? 'acryldata' : docker_registry
@@ -169,8 +169,8 @@ task runNoCode(type: Exec) {
 }
 
 docker {
+  dependsOn(bootJar)
   name "${docker_registry}/${docker_repo}:v${version}"
-  version "v${version}"
   dockerfile file("${rootProject.projectDir}/docker/${docker_repo}/Dockerfile")
   files bootJar.outputs.files
   files fileTree(rootProject.projectDir) {
@@ -181,12 +181,7 @@ docker {
   }.exclude {
     i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
   }
-  tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-  // platform('linux/arm64', 'linux/amd64')
-  buildx(true)
-  load(true)
-  push(false)
+  additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
   // Add build args if they are defined (needed for some CI or enterprise environments)
   def dockerBuildArgs = [:]
@@ -204,12 +199,4 @@ docker {
     buildArgs(dockerBuildArgs)
   }
 }
-tasks.getByPath(":datahub-upgrade:docker").dependsOn([bootJar])
-
-task cleanLocalDockerImages {
-  doLast {
-    rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-  }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)
 

datahub-web-react/.gitignore

@@ -28,3 +28,6 @@ yarn-error.log*
 # gql codegen
 *.generated.ts
 /.vscode
+
+.yarn-test-sentinel
+.yarn-lint-sentinel

datahub-web-react/build.gradle

@@ -80,6 +80,19 @@ task yarnServe(type: YarnTask, dependsOn: [yarnInstall, yarnGenerate]) {
 task yarnTest(type: YarnTask, dependsOn: [yarnInstall, yarnGenerate]) {
   // Explicitly runs in non-watch mode.
   args = ['run', project.hasProperty('withCoverage') ? 'test-coverage' : 'test', 'run']
+  def test_sentinel = "${buildDir}/.yarn-test-sentinel"
+  outputs.file(test_sentinel)
+  inputs.files(project.fileTree(dir: 'src', include: ['**/*.ts', '**/*.tsx']))
+  doLast {
+    // touch a file with name yarn-lint.txt in the build directory
+    def file = file(test_sentinel)
+    if (!file.exists()) {
+      file.createNewFile()
+    } else {
+      file.setLastModified(System.currentTimeMillis())
+    }
+  }
+  outputs.cacheIf { true }
 }
 
 task yarnLint(type: YarnTask, dependsOn: [yarnInstall, yarnGenerate]) {
@@ -90,6 +103,19 @@ test.dependsOn([yarnInstall, yarnTest, yarnLint])
 
 task yarnLintFix(type: YarnTask, dependsOn: [yarnInstall, yarnGenerate]) {
   args = ['run', 'lint-fix']
+  def lint_sentinel = "${buildDir}/.yarn-lint-sentinel"
+  outputs.file(lint_sentinel)
+  inputs.files(project.fileTree(dir: 'src', include: ['**/*.ts', '**/*.tsx']))
+  doLast {
+    // touch a file with name yarn-lint.txt in the build directory
+    def file = file(lint_sentinel)
+    if (!file.exists()) {
+      file.createNewFile()
+    } else {
+      file.setLastModified(System.currentTimeMillis())
+    }
+  }
+  outputs.cacheIf { true }
 }
 
 task yarnBuild(type: YarnTask, dependsOn: [yarnInstall, yarnGenerate]) {
@@ -130,9 +156,9 @@ distZip {
 }
 
 jar {
-  dependsOn distZip
+  dependsOn yarnBuild
   into('public') {
-    from zipTree(distZip.outputs.files.first())
+    from 'dist'
   }
   archiveClassifier = 'assets'
 }

docker/build.gradle

@@ -80,11 +80,25 @@ ext {
                     preserveVolumes: true
             ]
     ]
+
+    //  only for debug variants of quickstart to enable <variant>Reload tasks. 
+    // The actual service name needs the profile to be appended,  <container-name>-<profile>
+    moduleToContainer = [
+        ':metadata-service:war': 'datahub-gms',
+        ':datahub-frontend': 'frontend',
+        ':datahub-upgrade': 'system-update',
+        ':metadata-jobs:mce-consumer-job': 'datahub-mce-consumer',
+        ':metadata-jobs:mae-consumer-job': 'datahub-mae-consumer',
+
+    ]
 }
 
 // Register all quickstart tasks
 quickstart_configs.each { taskName, config ->
-    tasks.register(taskName)
+    tasks.register(taskName) {
+        group = 'quickstart'
+    }
+    
 }
 
 // Dynamically create all quickstart tasks and configurations
@@ -150,12 +164,57 @@ quickstart_configs.each { taskName, config ->
     }
 }
 
+// Register all quickstart tasks
+quickstart_configs.each { taskName, config ->
+    tasks.register("prepareAll${taskName}"){
+        group = 'quickstart-ci'
+    }
+}
+
+quickstart_configs.each { taskName, config ->
+    if (config.modules) {
+        tasks.getByName("prepareAll${taskName}").dependsOn(
+            config.modules.collect { it + ':dockerPrepare' }
+        )
+    }
+}
+
+quickstart_configs.each { taskName, config ->
+    tasks.register("buildImagesFromCache${taskName}") {
+        group = 'quickstart-ci'
+    }
+}
+
+quickstart_configs.each { taskName, config ->
+    if (config.modules) {
+        tasks.getByName("buildImagesFromCache${taskName}").dependsOn(
+            config.modules.collect { it + ':dockerFromCache' }
+        )
+    }
+}
+
+
+quickstart_configs.each { taskName, config ->
+    tasks.register("buildImages${taskName}") {
+        group = 'quickstart-ci'
+    }
+}
+
+quickstart_configs.each { taskName, config ->
+    if (config.modules) {
+        tasks.getByName("buildImages${taskName}").dependsOn(
+            config.modules.collect { it + ':dockerTag' }
+        )
+    }
+}
+
 tasks.register('minDockerCompose2.20', Exec) {
     executable 'bash'
     args '-c', 'echo -e "$(docker compose version --short)\n2.20"|sort --version-sort --check=quiet --reverse'
 }
 
 tasks.register('quickstartNuke') {
+    group = 'quickstart'
     doFirst {
         quickstart_configs.each { taskName, config ->
             dockerCompose."${taskName}".removeVolumes = !config.preserveVolumes
@@ -165,6 +224,7 @@ tasks.register('quickstartNuke') {
 }
 
 tasks.register('quickstartDown') {
+    group = 'quickstart'
     finalizedBy(tasks.withType(ComposeDownForced))
 }
 
@@ -173,12 +233,36 @@ tasks.withType(ComposeUp).configureEach {
     dependsOn tasks.named("minDockerCompose2.20")
 }
 
-task debugReload(type: Exec) {
-    def cmd = ['docker compose -p datahub --profile debug'] + ['-f', compose_base] + [
-            'restart',
-            'datahub-gms-debug',
-            'system-update-debug',
-            'frontend-debug'
-    ]
+// Register all quickstart Reload tasks. For quickstartDebug, the reload task is DebugReload. (Taskname without quickstart prefix)
+quickstart_configs.each { taskName, config ->
+    if (config.isDebug) {
+        def reloadTaskName = taskName.replaceFirst(/^quickstart/, "")
+        tasks.register("${reloadTaskName}Reload", Exec) {
+            dependsOn tasks.named("prepareAll${taskName}")
+            group = 'quickstart'
+            description = "Build and reload only changed containers for the ${taskName} task"
+            doFirst {
+                def executedTasks = project.gradle.taskGraph.allTasks.findAll { it.state.executed }
+                def containersToRestart = []
+
+                moduleToContainer.each { modulePath, containerName ->
+                    def moduleProject = project.project(modulePath)
+                    def dockerPrepareTask = moduleProject.tasks.findByName('dockerPrepare')
+                    
+                    if (dockerPrepareTask && executedTasks.contains(dockerPrepareTask) && !dockerPrepareTask.state.upToDate) {
+                        containersToRestart << "${containerName}-${config.profile}"
+                    }
+                }
+                
+                // Only restart containers that had their modules rebuilt
+                if (containersToRestart) {
+                    def cmd = ["docker compose -p datahub --profile ${config.profile}"] + ['-f', compose_base] + ['restart'] + containersToRestart
                     commandLine 'bash', '-c', cmd.join(" ")
+                } else {
+                    // If no containers need restart, make this a no-op
+                    commandLine 'bash', '-c', 'echo "No containers need restarting - all modules are up to date"'
+                }    
+            }
+        }
+    }
 }

docker/datahub-frontend/Dockerfile

@@ -25,17 +25,11 @@ RUN apk --no-cache --update-cache --available upgrade \
 
 ENV LD_LIBRARY_PATH="/lib:/lib64"
 
-FROM base AS unpack
-
-COPY ./datahub-frontend.zip /
-RUN unzip datahub-frontend.zip -d /tmp/out \
-   && mv /tmp/out/main /datahub-frontend
-COPY ./docker/monitoring/client-prometheus-config.yaml /datahub-frontend/
-RUN chown -R datahub:datahub /datahub-frontend && chmod 755 /datahub-frontend
-
 FROM base AS prod-install
 
-COPY --from=unpack /datahub-frontend/ /datahub-frontend/
+COPY --chown=datahub:datahub --chmod=755 ./docker/monitoring/client-prometheus-config.yaml /datahub-frontend/
+COPY --chown=datahub:datahub --chmod=755 ./docker/datahub-frontend/start.sh /
+COPY --chown=datahub:datahub --chmod=755 ./main /datahub-frontend/
 
 FROM base AS dev-install
 # Dummy stage for development. Assumes code is built on your machine and mounted to this image.
@@ -43,7 +37,6 @@ FROM base AS dev-install
 VOLUME [ "/datahub-frontend" ]
 
 FROM ${APP_ENV}-install AS final
-COPY --chown=datahub:datahub --chmod=755 ./docker/datahub-frontend/start.sh /
 USER datahub
 
 ENV OTEL_EXPORTER_OTLP_MAX_PAYLOAD_SIZE=4194304 \

docker/datahub-ingestion-base/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_registry = rootProject.ext.docker_registry == 'linkedin' ? 'acryldata' : docker_registry
@@ -16,8 +16,9 @@ ext {
 }
 
 docker {
+    dependsOn build
     name "${docker_registry}/${docker_repo}:v${docker_version}"
-    version "v${docker_version}"
+    //version "v${docker_version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -44,19 +45,3 @@ docker {
 
     buildArgs(dockerBuildArgs)
 }
-tasks.getByName('docker').dependsOn('build')
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-dockerClean.dependsOn([':docker:datahub-ingestion:dockerClean'])
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/datahub-ingestion/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_registry = rootProject.ext.docker_registry == 'linkedin' ? 'acryldata' : docker_registry
@@ -21,8 +21,8 @@ dependencies {
 }
 
 docker {
+    dependsOn 'build', ':docker:datahub-ingestion-base:docker', ':metadata-ingestion:codegen'
     name "${docker_registry}/${docker_repo}:v${docker_version}"
-    version "v${docker_version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile${docker_target == "slim" ? "-slim-only" : ""}")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -33,6 +33,7 @@ docker {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
 
+    version "v${docker_version}"
     def dockerBuildArgs = [DOCKER_VERSION: version, RELEASE_VERSION: version.replace('-SNAPSHOT', '').replace('v', '').replace("-slim", ''), BASE_IMAGE: "${docker_registry}/datahub-ingestion-base"]
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
@@ -45,20 +46,3 @@ docker {
 
     buildArgs(dockerBuildArgs)
 }
-tasks.getByName('dockerPrepare').dependsOn(['build',
-                                            ':docker:datahub-ingestion-base:docker',
-                                            ':metadata-ingestion:codegen'])
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/elasticsearch-setup/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_repo = 'datahub-elasticsearch-setup'
@@ -11,8 +11,8 @@ ext {
 }
 
 docker {
+    dependsOn(build)
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -21,12 +21,7 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-    // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -38,18 +33,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByName('docker').dependsOn('build')
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/kafka-setup/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_repo = 'datahub-kafka-setup'
@@ -11,8 +11,8 @@ ext {
 }
 
 docker {
+    dependsOn(build)
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -20,12 +20,7 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-    // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -46,18 +41,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByName('docker').dependsOn('build')
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/mysql-setup/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_registry = rootProject.ext.docker_registry == 'linkedin' ? 'acryldata' : docker_registry
@@ -12,8 +12,8 @@ ext {
 }
 
 docker {
+    dependsOn build
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -21,12 +21,7 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-    // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -38,18 +33,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByName('docker').dependsOn('build')
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/postgres-setup/build.gradle

@@ -1,9 +1,9 @@
 plugins {
-    id 'com.palantir.docker'
     id 'java' // required for versioning
 }
 
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_registry = rootProject.ext.docker_registry == 'linkedin' ? 'acryldata' : docker_registry
@@ -13,7 +13,6 @@ ext {
 
 docker {
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_dir}/Dockerfile")
     files fileTree(rootProject.projectDir) {
         include '.dockerignore'
@@ -21,12 +20,7 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-    // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -38,18 +32,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByName('docker').dependsOn('build')
-
-task mkdirBuildDocker {
-    doFirst {
-        mkdir "${project.buildDir}/docker"
-    }
-}
-dockerClean.finalizedBy(mkdirBuildDocker)
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

docker/profiles/docker-compose.frontend.yml

@@ -96,7 +96,7 @@ services:
     profiles:
       - debug-consumers
     depends_on:
-      system-update-debug:
+      system-update-debug-consumers:
         condition: service_completed_successfully
   frontend-debug-neo4j:
     <<: *datahub-frontend-service-dev

docker/profiles/docker-compose.gms.yml

@@ -248,12 +248,11 @@ services:
         condition: service_completed_successfully
       kafka-setup:
         condition: service_completed_successfully
-  system-update-debug:
+  system-update-debug: &system-update-debug
     <<: *datahub-system-update-service-dev
     profiles:
       - debug
       - debug-backend
-      - debug-consumers
     depends_on:
       mysql-setup-dev:
         condition: service_completed_successfully
@@ -261,6 +260,10 @@ services:
         condition: service_completed_successfully
       kafka-setup-dev:
         condition: service_completed_successfully
+  system-update-debug-consumers: # for consistent naming based on profile
+    <<: *system-update-debug
+    profiles:
+      - debug-consumers
   system-update-debug-elasticsearch:
     <<: *datahub-system-update-service-dev
     profiles:
@@ -386,7 +389,7 @@ services:
       MAE_CONSUMER_ENABLED: false
       MCE_CONSUMER_ENABLED: false
     depends_on:
-      system-update-debug:
+      system-update-debug-consumers:
         condition: service_completed_successfully
   datahub-gms-debug-neo4j:
     <<: *datahub-gms-service-dev
@@ -414,7 +417,7 @@ services:
     depends_on:
       datahub-gms-quickstart-consumers:
         condition: service_healthy
-  datahub-mae-consumer-quickstart-consumers-dev:
+  datahub-mae-consumer-debug-consumers:
     <<: *datahub-mae-consumer-service-dev
     profiles:
       - debug-consumers
@@ -431,7 +434,7 @@ services:
     depends_on:
       datahub-gms-quickstart-consumers:
         condition: service_healthy
-  datahub-mce-consumer-quickstart-consumers-dev:
+  datahub-mce-consumer-debug-consumers:
     <<: *datahub-mce-consumer-service-dev
     profiles:
       - debug-consumers

gradle/docker/docker-utils.gradle

@@ -0,0 +1,42 @@
+ext.getDockerImages = {
+  docker_registry, docker_repo, docker_tag ->
+    def stdOut = new ByteArrayOutputStream()
+    exec {
+      commandLine "docker", "images", "-q", "${docker_registry}/${docker_repo}:${docker_tag}"
+      standardOutput = stdOut
+    }
+    return stdOut.toString().trim().split("\\R").findAll {!it.empty}.unique() as List
+}
+
+ext.getDockerContainers = {
+  docker_registry, docker_repo, docker_tag ->
+    def stdOut = new ByteArrayOutputStream()
+    exec {
+      commandLine "docker", "container", "ls", "-q", "--filter", "ancestor=${docker_registry}/${docker_repo}:${docker_tag}"
+      standardOutput = stdOut
+    }
+    return stdOut.toString().trim().split("\\R").findAll {!it.empty}.unique() as List
+}
+
+ext.cleanLocalDockerImages = {
+  String docker_registry, String docker_repo, String docker_tag ->
+    println("Docker image string: ${docker_registry}/${docker_repo}:${docker_tag}")
+    def containers = getDockerContainers(docker_registry, docker_repo, docker_tag)
+    if(!containers.isEmpty()) {
+      println "Stopping containers: $containers"
+      exec {
+        commandLine = ["docker", "container", "stop"] + containers
+      }
+      exec {
+        commandLine = ["docker", "container", "rm"] + containers
+      }
+    }
+    def images = getDockerImages(docker_registry, docker_repo, docker_tag)
+    if(!images.isEmpty()) {
+      println "Removing images: $images"
+      exec {
+        ignoreExitValue true // may not work if used by downstream image
+        commandLine = ["docker", "rmi", "-f"] + images
+      }
+    }
+}

gradle/docker/docker.gradle

@@ -1,29 +1,30 @@
+import groovy.json.JsonSlurper
+import org.apache.commons.io.output.TeeOutputStream
 
-ext.getDockerImages = {
-    docker_registry, docker_repo, docker_tag ->
+
+def _getDockerImages(String fullImageTag) {
   def stdOut = new ByteArrayOutputStream()
   exec {
-            commandLine "docker", "images", "-q", "${docker_registry}/${docker_repo}:${docker_tag}"
+    commandLine "docker", "images", "-q", "${fullImageTag}"
     standardOutput = stdOut
   }
-        return stdOut.toString().trim().split("\\R").findAll {!it.empty}.unique() as List
+  return stdOut.toString().trim().split("\\R").findAll { !it.empty }.unique() as List
 }
 
-ext.getDockerContainers = {
-    docker_registry, docker_repo, docker_tag ->
+def _getDockerContainers(String fullImageTag) {
   def stdOut = new ByteArrayOutputStream()
   exec {
-            commandLine "docker", "container", "ls", "-q", "--filter", "ancestor=${docker_registry}/${docker_repo}:${docker_tag}"
+    commandLine "docker", "container", "ls", "-q", "--filter",
+        "ancestor=${fullImageTag}"
     standardOutput = stdOut
   }
-        return stdOut.toString().trim().split("\\R").findAll {!it.empty}.unique() as List
+  return stdOut.toString().trim().split("\\R").findAll { !it.empty }.unique() as List
 }
 
-ext.cleanLocalDockerImages = {
-    String docker_registry, String docker_repo, String docker_tag ->
-        println("Docker image string: ${docker_registry}/${docker_repo}:${docker_tag}")
-        def containers = getDockerContainers(docker_registry, docker_repo, docker_tag)
-        if(!containers.isEmpty()) {
+def _cleanLocalDockerImages(String fullImageTag) {
+  println("Docker image string: ${fullImageTag}")
+  def containers = _getDockerContainers(fullImageTag)
+  if (!containers.isEmpty()) {
     println "Stopping containers: $containers"
     exec {
       commandLine = ["docker", "container", "stop"] + containers
@@ -32,8 +33,8 @@ ext.cleanLocalDockerImages = {
       commandLine = ["docker", "container", "rm"] + containers
     }
   }
-        def images = getDockerImages(docker_registry, docker_repo, docker_tag)
-        if(!images.isEmpty()) {
+  def images = _getDockerImages(fullImageTag)
+  if (!images.isEmpty()) {
     println "Removing images: $images"
     exec {
       ignoreExitValue true // may not work if used by downstream image
@@ -41,3 +42,191 @@ ext.cleanLocalDockerImages = {
     }
   }
 }
+
+// Create extension object
+class DockerPluginExtension {
+  Project project
+  Property<File> dockerfile
+  CopySpec copySpec
+  MapProperty<String, String> buildArgs
+  MapProperty<String, String> tags
+  ListProperty<String> platforms
+  ListProperty<Object> dependencies // May contain tasks or task  names
+
+  DockerPluginExtension(Project project) {
+    this.project = project
+    dockerfile = project.objects.property(File)
+    buildArgs = project.objects.mapProperty(String, String)
+    copySpec = project.copySpec()
+    tags = project.objects.mapProperty(String, String)
+    platforms = project.objects.listProperty(String)
+    dependencies = project.objects.listProperty(Object)
+  }
+
+  def files(Object... files) {
+    copySpec.from(files)
+  }
+
+  def name(String value) {
+    additionalTag("", value)
+  }
+
+  def dockerfile(File value) {
+    dockerfile.set(value)
+  }
+
+  def buildArgs(Map<String, String> values) {
+    buildArgs.putAll(values)
+  }
+
+  def platform(String... platforms) {
+    this.platforms.addAll(platforms)
+  }
+
+  def additionalTag(String name, String tag) {
+    tags.put(name, tag)
+  }
+
+  def dependsOn(Object... tasks) {
+    dependencies.addAll(tasks)
+  }
+}
+
+def extension = project.extensions.create("docker", DockerPluginExtension)
+
+project.afterEvaluate {
+  def buildContext = "${rootProject.buildDir}/dockerBuildContext/${rootProject.relativePath(project.projectDir)}/docker"
+  // ensure this directory exists
+  new File(buildContext).mkdirs()
+  println("buildContext: ${buildContext}")
+
+  tasks.register("dockerPrepare", Sync) {
+    group "docker"
+    with extension.copySpec
+    from extension.dockerfile
+    into buildContext
+    dependsOn extension.dependencies.get()
+  }
+
+  project.tasks.register("docker", Exec) {
+    group "docker"
+    description "Builds the docker image and applies all tags defined"
+    dependsOn dockerPrepare
+
+    def marker = "${buildContext}/../imageCreated-${name}.json"
+
+    inputs.file(extension.dockerfile)
+    inputs.dir(buildContext)
+    inputs.property("tags", extension.tags)
+    inputs.property("buildArgs", extension.buildArgs)
+    outputs.file(marker)
+
+    def dockerCmd = []
+
+    dockerCmd += ["docker", "buildx", "build", "--load"]
+    if (extension.platforms.get()) {
+      dockerCmd << "--platform=${extension.platforms.get().join(',')}"
+    }
+
+    // Add GitHub Actions specific arguments if running in GitHub workflow and using gha cache.
+    // Or set DOCKER_CACHE=DEPOT -- which transparently can cache without these args when running on depot runner.
+    if (System.getenv("DOCKER_CACHE") == "GITHUB") {
+      def githubToken = System.getenv("GITHUB_TOKEN")
+      if (githubToken) {
+        dockerCmd += ["--cache-from", "type=gha,token=${githubToken}"]
+        dockerCmd += ["--cache-to", "type=gha,mode=max,token=${githubToken}"]
+      } else {
+        dockerCmd += ["--cache-from", "type=gha"]
+        dockerCmd += ["--cache-to", "type=gha,mode=max"]
+      }
+    }
+
+    // Generate image metadata (we really just want the sha256 hash of the image)
+    dockerCmd += ["--metadata-file", marker]
+
+    extension.buildArgs.get().each { k, v -> dockerCmd += ["--build-arg", "${k}=${v}"]
+    }
+
+    extension.tags.get().each { taskName, tag ->
+      dockerCmd += ["-t", tag]
+    }
+
+    dockerCmd << buildContext
+
+    // Some projects use a Dockerfile with the non-default name.
+    dockerCmd += ["--file", extension.dockerfile.get().toPath()]
+
+    // Mask GitHub token in the output for security
+    def maskedCmd = dockerCmd.collect { arg ->
+      if (arg.startsWith("type=gha,token=")) {
+        "type=gha,token=****"
+      } else {
+        arg
+      }
+    }
+    println(maskedCmd.join(" "))
+    commandLine dockerCmd
+
+    outputs.upToDateWhen {
+      try {
+        /* The docker task is up-to-date if
+         * 1. the last build generated a marker file
+         * 2. An image with the same tag exists in local docker images
+         * 3. that existing image sha256 matches what is written in the generated marker file
+         */
+        def jsonContent = new File(marker).text
+        def jsonData = new JsonSlurper().parseText(jsonContent)
+        def imageIdFromMarker = jsonData['containerimage.digest']
+        if (imageIdFromMarker != null && imageIdFromMarker.startsWith("sha256:")) {
+          imageIdFromMarker = imageIdFromMarker.substring(7); // "sha256:".length() == 7
+        }
+        for(String tag : extension.tags.get().values()) {
+          def actualImage = _getDockerImages(tag)
+          if (actualImage == null || actualImage.size() == 0 && imageIdFromMarker == null ||
+              !imageIdFromMarker.startsWith(actualImage.get(0))) {
+            logger.debug("UP-TO-DATE CHECK for ${name}: did not find image ${imageIdFromMarker}")
+            return false
+          }
+        }
+        logger.debug("UP-TO-DATE CHECK for ${name}: Is up-to-date, skipping")
+        return true
+      } catch (Exception e) {
+        // any exceptions also implicitly mean not-up-to-date
+        return false
+      }
+    }
+  }
+
+  project.tasks.register("dockerFromCache") {
+    // This task is same as docker but without the dockerPrepare dependency. This is useful in CI where a
+    // github workflow can run multiple jobs in parallel which all share the docker build context root folder.
+    // This is faster than uploading and downloading all images together.
+    group "docker"
+    description "Builds the docker image from cache and applies all tags defined but without the dependencies."
+    // Reference the docker original task's configuration and run it but without its dependencies
+    doLast {
+      def originalTaskInstance = tasks.named('docker').get()
+      project.exec {
+        commandLine originalTaskInstance.commandLine
+        workingDir originalTaskInstance.workingDir
+      }
+    }
+  }
+
+  extension.tags.get().each { taskName, tag ->
+    // For backward compatibility, can be removed if we dont really have a need post migration
+    // TODO: Choice of task names is to retain current names so that downstream dependencies in quickstart still work
+    //  without changes. Can be changed post full migration.
+    project.tasks.register("dockerTag${taskName}") {
+      dependsOn project.tasks.named("docker")
+    }
+  }
+
+  task dockerClean {
+    group "docker"
+    doLast {
+      extension.tags.get().each { _, tag -> _cleanLocalDockerImages(tag)
+      }
+    }
+  }
+}

metadata-integration/java/spark-lineage-legacy/build.gradle

@@ -230,11 +230,11 @@ nexusStaging {
     password = System.getenv("NEXUS_PASSWORD")
 }
 
-task cleanExtraDirs {
+clean{
   delete "$projectDir/derby.log"
   delete "$projectDir/src/test/resources/data/hive"
   delete "$projectDir/src/test/resources/data/out.csv"
   delete "$projectDir/src/test/resources/data/out_persist.csv"
   delete "$projectDir/spark-smoke-test/venv"
 }
-clean.finalizedBy(cleanExtraDirs)
+

metadata-jobs/mae-consumer-job/build.gradle

@@ -1,11 +1,11 @@
 plugins {
     id 'org.springframework.boot'
     id 'java'
-    id 'com.palantir.docker'
 }
 
 apply from: '../../gradle/coverage/java-coverage.gradle'
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_repo = 'datahub-mae-consumer'
@@ -44,8 +44,9 @@ bootJar {
 }
 
 docker {
+    dependsOn(bootJar)
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
+    //version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_repo}/Dockerfile")
     files bootJar.outputs.files
     files fileTree(rootProject.projectDir) {
@@ -56,12 +57,9 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -79,11 +77,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByPath(":metadata-jobs:mae-consumer-job:docker").dependsOn([bootJar])
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

metadata-jobs/mce-consumer-job/build.gradle

@@ -1,11 +1,11 @@
 plugins {
     id 'org.springframework.boot'
     id 'java'
-    id 'com.palantir.docker'
 }
 
 apply from: '../../gradle/coverage/java-coverage.gradle'
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
     docker_repo = 'datahub-mce-consumer'
@@ -55,8 +55,9 @@ bootJar {
 }
 
 docker {
+    dependsOn(bootJar)
     name "${docker_registry}/${docker_repo}:v${version}"
-    version "v${version}"
+    //version "v${version}"
     dockerfile file("${rootProject.projectDir}/docker/${docker_repo}/Dockerfile")
     files bootJar.outputs.files
     files fileTree(rootProject.projectDir) {
@@ -67,12 +68,9 @@ docker {
     }.exclude {
         i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
     }
-    tag("Debug", "${docker_registry}/${docker_repo}:debug")
+    additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
     // platform('linux/arm64', 'linux/amd64')
-    buildx(true)
-    load(true)
-    push(false)
 
     // Add build args if they are defined (needed for some CI or enterprise environments)
     def dockerBuildArgs = [:]
@@ -90,11 +88,3 @@ docker {
         buildArgs(dockerBuildArgs)
     }
 }
-tasks.getByPath(":metadata-jobs:mce-consumer-job:docker").dependsOn([bootJar])
-
-task cleanLocalDockerImages {
-    doLast {
-        rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-    }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)

metadata-service/war/build.gradle

@@ -1,11 +1,11 @@
 plugins {
   id 'org.springframework.boot'
-  id 'com.palantir.docker'
   id 'java'
 }
 
 apply from: '../../gradle/coverage/java-coverage.gradle'
 apply from: "../../gradle/versioning/versioning.gradle"
+apply from: "../../gradle/docker/docker.gradle"
 
 ext {
   docker_repo = 'datahub-gms'
@@ -83,8 +83,8 @@ bootRun {
 }
 
 docker {
+  dependsOn bootJar
   name "${docker_registry}/${docker_repo}:v${version}"
-  version "v${version}"
   dockerfile file("${rootProject.projectDir}/docker/${docker_repo}/Dockerfile")
   files bootJar.outputs.files
   files fileTree(rootProject.projectDir) {
@@ -95,12 +95,7 @@ docker {
   }.exclude {
     i -> (!i.file.name.endsWith(".dockerignore") && i.file.isHidden())
   }
-  tag("Debug", "${docker_registry}/${docker_repo}:debug")
-
-  // platform('linux/arm64', 'linux/amd64')
-  buildx(true)
-  load(true)
-  push(false)
+  additionalTag("Debug", "${docker_registry}/${docker_repo}:debug")
 
   // Add build args if they are defined (needed for some CI or enterprise environments)
   def dockerBuildArgs = [:]
@@ -118,14 +113,6 @@ docker {
     buildArgs(dockerBuildArgs)
   }
 }
-tasks.getByPath(":metadata-service:war:docker").dependsOn([bootJar])
-
-task cleanLocalDockerImages {
-  doLast {
-    rootProject.ext.cleanLocalDockerImages(docker_registry, docker_repo, "${version}")
-  }
-}
-dockerClean.finalizedBy(cleanLocalDockerImages)
 
 test {
   jacoco {