Cache Gradle distributions with MASS fallback

Author: AlexeyKuznetsov-DD

.gitlab-ci.yml

@@ -251,11 +251,6 @@ default:
     # Apache Maven Wrapper supports MVNW_REPOURL for repository-manager downloads:
     # https://maven.apache.org/tools/wrapper/#Using_a_Maven_Repository_Manager
     - export MVNW_REPOURL=${MAVEN_REPOSITORY_PROXY%/}
-    # Route Gradle distribution download through MASS pull-through cache
-    - |
-      mass_read_host="${MASS_READ_URL#https://}"
-      mass_read_host="${mass_read_host%/}"
-      sed -i "/^distributionUrl=/ s|services.gradle.org|${mass_read_host}/internal/artifact/services.gradle.org|" gradle/wrapper/gradle-wrapper.properties
     - mkdir -p .mvn/caches
     # Redirect Spotless's Equo/Solstice P2 cache into the project tree so it is captured by the GitLab cache.
     # Solstice (https://github.com/equodev/equo-ide) defaults to ~/.m2/repository/dev/equo/p2-data, which is outside $CI_PROJECT_DIR.
@@ -271,6 +266,8 @@ default:
     # GitLab's cache helper restores .gradle as root, but we run as non-root-user (uid 1001),
     # and Gradle does `chmod 700 .gradle` on startup which requires user ownership.
     - sudo chown -R 1001:1001 .gradle
+    # Bootstrap Gradle through MASS with upstream fallback.
+    - bootstrap_gradle_distribution
   after_script:
     - *cgroup_info
     - *container_info
@@ -818,6 +815,7 @@ muzzle-dep-report:
         - dependency-lib
       unprotect: true
   before_script:
+    - source .gitlab/gitlab-utils.sh
     - git config --global --add safe.directory "$CI_PROJECT_DIR"
     # Akka token added to SSM from https://account.akka.io/token
     - export ORG_GRADLE_PROJECT_akkaRepositoryToken=$(aws ssm get-parameter --region us-east-1 --name ci.dd-trace-java.akka_repo_token --with-decryption --query "Parameter.Value" --out text)
@@ -838,17 +836,13 @@ muzzle-dep-report:
     # Apache Maven Wrapper supports MVNW_REPOURL for repository-manager downloads:
     # https://maven.apache.org/tools/wrapper/#Using_a_Maven_Repository_Manager
     - export MVNW_REPOURL=${MAVEN_REPOSITORY_PROXY%/}
-    # Route Gradle distribution download through MASS pull-through cache
-    - |
-      mass_read_host="${MASS_READ_URL#https://}"
-      mass_read_host="${mass_read_host%/}"
-      sed -i "/^distributionUrl=/ s|services.gradle.org|${mass_read_host}/internal/artifact/services.gradle.org|" gradle/wrapper/gradle-wrapper.properties
     - *normalize_node_index
     - *prepare_test_env
     # Disable CDS in forked JVMs to avoid SIGSEGVs on Linux arm64.
     - export GRADLE_OPTS="-Dorg.gradle.jvmargs='-Xshare:off -Xms$GRADLE_MEMORY_MIN -Xmx$GRADLE_MEMORY_MAX -XX:ErrorFile=/tmp/hs_err_pid%p.log -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/tmp -Djava.util.prefs.userRoot=/tmp/.java/.userPrefs-${CI_JOB_ID}' -Ddatadog.forkedMinHeapSize=128M -Ddatadog.forkedMaxHeapSize=1024M"
     - export GRADLE_ARGS="--build-cache --stacktrace --no-daemon --parallel --max-workers=$GRADLE_WORKERS"
-    - ./gradlew --version
+    # Bootstrap Gradle through MASS with upstream fallback.
+    - bootstrap_gradle_distribution
   script:
     - *gitlab_base_ref_params
     - ./gradlew $GRADLE_TARGET $GRADLE_PARAMS -PtestJvm=$testJvm -Pslot=$CI_NODE_INDEX/$CI_NODE_TOTAL $GRADLE_ARGS --continue

.gitlab/gitlab-utils.sh

@@ -16,6 +16,51 @@ function gitlab_section_end () {
   echo -e "section_end:`date +%s`:${section_title}\r\e[0K"
 }
 
+# Bootstrap Gradle through MASS, retrying upstream because the wrapper supports only one URL.
+# The pinned checksum applies to either source. Call after `.gradle` and the Gradle environment are
+# prepared.
+function bootstrap_gradle_distribution () {
+  local props="gradle/wrapper/gradle-wrapper.properties"
+  local upstream_props="/tmp/gradle-wrapper.upstream.properties"
+  local mass_read_host
+
+  if [ -z "${MASS_READ_URL:-}" ]; then
+    ./gradlew --version
+    return
+  fi
+
+  mass_read_host="${MASS_READ_URL#https://}"
+  mass_read_host="${mass_read_host%/}"
+
+  # Derive upstream even if an earlier call already routed the URL through MASS.
+  sed "/^distributionUrl=/ s|${mass_read_host}/internal/artifact/||" "$props" > "$upstream_props"
+
+  if grep -q "^distributionUrl=.*${mass_read_host}" "$props"; then
+    # Avoid nesting the MASS prefix.
+    echo "Gradle distribution is already routed through ${mass_read_host}"
+  else
+    # Redirect because GNU and BSD `sed -i` differ.
+    sed "/^distributionUrl=/ s|services.gradle.org|${mass_read_host}/internal/artifact/services.gradle.org|" "$props" > "$props.mass" &&
+      mv "$props.mass" "$props"
+  fi
+
+  if ./gradlew --version; then
+    rm -f "$upstream_props"
+    return
+  fi
+
+  if ! grep -q "^distributionUrl=" "$upstream_props"; then
+    echo -e "${TEXT_RED}Gradle distribution bootstrap failed and no upstream distributionUrl could be derived from ${props} to fall back to${TEXT_CLEAR}" >&2
+    return 1
+  fi
+
+  # Keep upstream: changing the URL changes the wrapper cache key and would download twice.
+  echo -e "${TEXT_YELLOW}MASS_FALLBACK gradle-distribution: ${MASS_READ_URL} could not serve the Gradle distribution, retrying via services.gradle.org${TEXT_CLEAR}" >&2
+  cp "$upstream_props" "$props"
+  rm -f "$upstream_props"
+  ./gradlew --version
+}
+
 # A subset of ansi color/formatting codes https://misc.flogisoft.com/bash/tip_colors_and_formatting
 export TEXT_RED="\e[31m"
 export TEXT_GREEN="\e[32m"

build-logic/conventions/build.gradle.kts

@@ -1,5 +1,6 @@
 plugins {
   `kotlin-dsl`
+  `jvm-test-suite`
 }
 
 java {
@@ -12,3 +13,17 @@ kotlin {
     jvmTarget.set(org.jetbrains.kotlin.gradle.dsl.JvmTarget.JVM_1_8)
   }
 }
+
+@Suppress("UnstableApiUsage")
+testing {
+  suites {
+    named<JvmTestSuite>("test") {
+      useJUnitJupiter(libs.versions.junit5)
+      dependencies {
+        implementation(libs.junit.jupiter)
+        implementation(libs.junit.jupiter.engine)
+        implementation(libs.assertj.core)
+      }
+    }
+  }
+}

build-logic/conventions/src/main/kotlin/datadog/buildlogic/mass/MassExtension.kt

@@ -11,9 +11,16 @@ constructor(objects: ObjectFactory, providers: ProviderFactory) {
   val readUrl: Property<String> =
     objects.property(String::class.java).convention(providers.environmentVariable("MASS_READ_URL"))
 
-  fun artifactUrl(upstreamArtifactUrl: String): String {
-    val massReadUrl = readUrl.orNull ?: return "https://$upstreamArtifactUrl"
+  fun artifactUrl(upstreamArtifactUrl: String): String = artifactUrls(upstreamArtifactUrl).first()
+
+  /**
+   * Ordered repository URLs: MASS, then upstream. Gradle advances only for missing artifacts, so
+   * this covers MASS cache misses, not connection failures. Without MASS, returns only upstream.
+   */
+  fun artifactUrls(upstreamArtifactUrl: String): List<String> {
+    val upstreamUrl = "https://$upstreamArtifactUrl"
+    val massReadUrl = readUrl.orNull?.takeIf { it.isNotBlank() } ?: return listOf(upstreamUrl)
     val baseUrl = if (massReadUrl.endsWith("/")) massReadUrl else "$massReadUrl/"
-    return "${baseUrl}internal/artifact/$upstreamArtifactUrl"
+    return listOf("${baseUrl}internal/artifact/$upstreamArtifactUrl", upstreamUrl)
   }
 }

build-logic/conventions/src/test/kotlin/datadog/buildlogic/mass/MassExtensionTest.kt

@@ -0,0 +1,56 @@
+package datadog.buildlogic.mass
+
+import org.assertj.core.api.Assertions.assertThat
+import org.gradle.kotlin.dsl.newInstance
+import org.gradle.testfixtures.ProjectBuilder
+import org.junit.jupiter.api.Test
+
+class MassExtensionTest {
+
+  private fun extension(massReadUrl: String?): MassExtension {
+    val project = ProjectBuilder.builder().build()
+    // Empty overrides the environment-backed convention.
+    return project.objects.newInstance<MassExtension>().also { it.readUrl.set(massReadUrl ?: "") }
+  }
+
+  @Test
+  fun `routes artifacts through MASS and keeps the upstream host as a second repository`() {
+    val mass = extension("https://mass.example")
+
+    assertThat(mass.artifactUrls("dlcdn.apache.org")).containsExactly(
+      "https://mass.example/internal/artifact/dlcdn.apache.org",
+      "https://dlcdn.apache.org",
+    )
+    assertThat(mass.artifactUrl("dlcdn.apache.org"))
+      .isEqualTo("https://mass.example/internal/artifact/dlcdn.apache.org")
+  }
+
+  @Test
+  fun `tolerates a trailing slash on the MASS read url`() {
+    val mass = extension("https://mass.example/")
+
+    assertThat(mass.artifactUrls("dlcdn.apache.org").first())
+      .isEqualTo("https://mass.example/internal/artifact/dlcdn.apache.org")
+  }
+
+  @Test
+  fun `declares the upstream host only once when MASS is not configured`() {
+    listOf(null, "", "   ").forEach { unset ->
+      val mass = extension(unset)
+
+      assertThat(mass.artifactUrls("dlcdn.apache.org"))
+        .containsExactly("https://dlcdn.apache.org")
+      assertThat(mass.artifactUrl("dlcdn.apache.org")).isEqualTo("https://dlcdn.apache.org")
+    }
+  }
+
+  @Test
+  fun `keeps any path on the upstream artifact url`() {
+    val mass = extension("https://mass.example")
+
+    assertThat(mass.artifactUrls("github.com/wildfly/wildfly/releases/download/")).containsExactly(
+      "https://mass.example/internal/artifact/github.com/wildfly/wildfly/releases/download/",
+      "https://github.com/wildfly/wildfly/releases/download/",
+    )
+  }
+}

build-logic/smoke-test/src/main/kotlin/datadog/buildlogic/smoketest/GradleDistribution.kt

@@ -4,9 +4,26 @@ import java.net.URI
 
 internal const val MASS_READ_URL_ENV = "MASS_READ_URL"
 
+/** Public Gradle distribution source. */
+internal const val UPSTREAM_DISTRIBUTIONS_BASE_URL = "https://services.gradle.org/distributions"
+
 internal fun gradleDistributionUri(massReadUrl: String, gradleVersion: String): URI {
   val baseUrl = if (massReadUrl.endsWith("/")) massReadUrl else "$massReadUrl/"
   return URI.create(
     "${baseUrl}internal/artifact/services.gradle.org/distributions/gradle-$gradleVersion-bin.zip",
   )
 }
+
+internal fun upstreamGradleDistributionUri(gradleVersion: String): URI =
+  URI.create("$UPSTREAM_DISTRIBUTIONS_BASE_URL/gradle-$gradleVersion-bin.zip")
+
+/** Nested Gradle sources, preferring MASS and falling back upstream. */
+internal fun gradleDistributionUris(massReadUrl: String?, gradleVersion: String): List<URI> =
+  if (massReadUrl.isNullOrBlank()) {
+    listOf(upstreamGradleDistributionUri(gradleVersion))
+  } else {
+    listOf(
+      gradleDistributionUri(massReadUrl, gradleVersion),
+      upstreamGradleDistributionUri(gradleVersion),
+    )
+  }