Copilot commented on code in PR #12278: URL: https://github.com/apache/gluten/pull/12278#discussion_r3415663329
########## .github/workflows/delta_spark_ut.yml: ########## @@ -0,0 +1,656 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# Runs Delta Lake's `spark` sbt module unit tests against a Gluten Velox bundle +# that is built from the source in this repository. The pipeline: +# +# 1. Builds the Velox/Gluten native libraries (centos-7 + vcpkg, x86_64). +# 2. Builds the Gluten Java/Scala jars and assembles the +# `gluten-velox-bundle-spark<spark>_<scala>-linux_amd64-<version>.jar` +# fat jar for Spark 4.1 + Scala 2.13 + Java 17 with the Delta profile. +# 3. Clones delta-io/delta at the requested release tag (default `v4.2.0`), +# drops the bundle jar into `spark-unified/lib/` only (NOT `spark/lib/` +# -- see setup-delta.sh for the unmanagedJars scoping rationale), +# patches Delta's `DeltaSQLCommandTest` to register the Gluten plugin, +# and runs `sbt spark/test` sharded across the matrix. +# +# Limited to Velox + x86 to keep the matrix simple, per the pipeline's purpose +# of validating Gluten changes against the latest Delta release. + +name: Delta Spark UT (Gluten) + +on: + workflow_dispatch: + inputs: + delta_ref: + description: 'delta-io/delta git ref (tag/branch/SHA) to test against' + required: true + default: 'v4.2.0' + spark_version: + description: 'Delta `-DsparkVersion` value (must match the Gluten -P profile below)' + required: true + default: '4.1' + test_parallelism: + description: 'Forked test JVMs per shard (TEST_PARALLELISM_COUNT)' + required: true + default: '1' + update_baseline: + description: 'Seed/refresh the known-failures baseline instead of enforcing it' + type: boolean + required: false + default: false + fail_on_fixed: + description: 'Fail when a baseline test now passes (keeps the baseline honest)' + type: boolean + required: false + default: true + pull_request: + paths: + - '.github/workflows/delta_spark_ut.yml' + - '.github/workflows/util/delta-spark-ut/**' + - 'gluten-delta/**' + - 'backends-velox/src-delta40/**/DeltaSQLCommandTest.scala' + +env: + ACTIONS_ALLOW_USE_UNSECURE_NODE_VERSION: true + MVN_CMD: 'build/mvn -ntp' + CCACHE_DIR: "${{ github.workspace }}/.ccache" + # Gluten profile / bundle naming for the build-gluten-bundle and + # delta-spark-test jobs. Spark 4.1 + Scala 2.13 + JDK 17 matches Delta v4.2.0's + # default Spark version (4.1.0) from project/CrossSparkVersions.scala. + GLUTEN_SPARK_PROFILE: 'spark-4.1' + GLUTEN_SCALA_PROFILE: 'scala-2.13' + GLUTEN_JAVA_PROFILE: 'java-17' + GLUTEN_BUNDLE_SPARK_VERSION: '4.1' + GLUTEN_BUNDLE_SCALA_VERSION: '2.13' + # Default values used when the workflow is triggered by pull_request + # (where `inputs.*` is empty). Keep these in sync with the workflow_dispatch + # defaults above. + DELTA_REF_DEFAULT: 'v4.2.0' + DELTA_SPARK_VERSION_DEFAULT: '4.1' + DELTA_TEST_PARALLELISM_DEFAULT: '1' + # Default mode for pull_request runs (where inputs.* is empty): enforce the + # committed baseline and fail when a baseline test starts passing. Override + # via the workflow_dispatch inputs above. + DELTA_UPDATE_BASELINE_DEFAULT: 'false' + DELTA_FAIL_ON_FIXED_DEFAULT: 'true' + DELTA_SCALA_VERSION: '2.13.16' + # Number of shards in the delta-spark-test matrix. Must equal the length of + # the `shard` matrix below. + # + # Set to 16 (not 8): at 8-way, ~5 of the 8 shards consistently exceed the + # 300-min job timeout because Delta's generated merge/CDC/DV suites are very + # slow under Gluten (observed in run 16: shards 0,1,2,3,6 all hit the 300-min + # cap while 4,5,7 finished in <130 min). Delta's GreedyHashStrategy balances + # high-duration suites across shards by estimated duration, so doubling the + # shard count roughly halves per-shard wall time and lets every shard finish + # (and thus report its results) within the timeout. We keep + # TEST_PARALLELISM_COUNT=1 because each forked test JVM uses ~10G (8G heap + + # 2G off-heap) and >1 would OOM the runner. + DELTA_NUM_SHARDS: '16' + +concurrency: + group: ${{ github.repository }}-${{ github.head_ref || github.sha }}-${{ github.workflow }} + cancel-in-progress: true + +jobs: + build-native-lib-centos-7: + runs-on: ubuntu-22.04 + steps: + - uses: actions/checkout@v4 + - name: Get Ccache + uses: actions/cache/restore@v4 + with: + path: '${{ env.CCACHE_DIR }}' + key: ccache-delta-spark-ut-centos7-release-default-${{github.sha}} + restore-keys: | + ccache-delta-spark-ut-centos7-release-default + ccache-centos7-release-default + - name: Build Gluten native libraries + run: | + docker run -v $GITHUB_WORKSPACE:/work -w /work apache/gluten:vcpkg-centos-7-gcc13 bash -c " + set -e + yum install tzdata -y + df -a + cd /work + export CCACHE_DIR=/work/.ccache + export CCACHE_MAXSIZE=1G + mkdir -p /work/.ccache + ccache -sz + bash dev/ci-velox-buildstatic-centos-7.sh + ccache -s + mkdir -p /work/.m2/repository/org/apache/arrow/ + cp -r /root/.m2/repository/org/apache/arrow/* /work/.m2/repository/org/apache/arrow/ + " + - name: Save Ccache + if: always() + uses: actions/cache/save@v4 + with: + path: '${{ env.CCACHE_DIR }}' + key: ccache-delta-spark-ut-centos7-release-default-${{github.sha}} + - uses: actions/upload-artifact@v4 + with: + name: delta-spark-ut-native-lib-centos-7-${{github.sha}} + path: ./cpp/build/ + if-no-files-found: error + - uses: actions/upload-artifact@v4 + with: + name: delta-spark-ut-arrow-jars-centos-7-${{github.sha}} + path: .m2/repository/org/apache/arrow/ + if-no-files-found: error + + build-gluten-bundle: + needs: build-native-lib-centos-7 + runs-on: ubuntu-22.04 + container: apache/gluten:centos-9-jdk17 + steps: + - uses: actions/checkout@v4 + - name: Download native artifacts + uses: actions/download-artifact@v4 + with: + name: delta-spark-ut-native-lib-centos-7-${{github.sha}} + path: ./cpp/build/ + - name: Download Arrow jars + uses: actions/download-artifact@v4 + with: + name: delta-spark-ut-arrow-jars-centos-7-${{github.sha}} + path: /root/.m2/repository/org/apache/arrow/ + - name: Cache Maven repository + uses: actions/cache@v4 + with: + path: /root/.m2/repository + key: m2-delta-spark-ut-bundle-${{ env.GLUTEN_SPARK_PROFILE }}-${{ env.GLUTEN_SCALA_PROFILE }}-${{ hashFiles('pom.xml', '**/pom.xml') }} + restore-keys: | + m2-delta-spark-ut-bundle-${{ env.GLUTEN_SPARK_PROFILE }}-${{ env.GLUTEN_SCALA_PROFILE }}- + m2-delta-spark-ut-bundle- + - name: Build Gluten Velox + Delta bundle + run: | + set -euo pipefail + yum install -y java-17-openjdk-devel + export JAVA_HOME=/usr/lib/jvm/java-17-openjdk + export PATH=$JAVA_HOME/bin:$PATH + java -version + cd "$GITHUB_WORKSPACE" + # `install` (not `package`) so the gluten-delta artifact is in the local + # m2 repo before the `package/` shaded jar is built. `Dmaven.compiler.release=17` + # overrides any user settings.xml that may pin release=1.8 for Java 17 builds. + $MVN_CMD clean install \ + -P${{ env.GLUTEN_SPARK_PROFILE }} \ + -P${{ env.GLUTEN_SCALA_PROFILE }} \ + -P${{ env.GLUTEN_JAVA_PROFILE }} \ + -Pbackends-velox -Pdelta \ + -DskipTests -Dmaven.compiler.release=17 + - name: Stage bundle jar + run: | + set -euo pipefail + mkdir -p bundle-out + # Match the renamed fat jar produced by package/pom.xml's copy-fat-jar + # exec. The version part may bump (e.g. 1.7.0-SNAPSHOT -> 1.8.0-SNAPSHOT), + # so glob the version suffix. + jar=$(ls package/target/gluten-velox-bundle-spark${{ env.GLUTEN_BUNDLE_SPARK_VERSION }}_${{ env.GLUTEN_BUNDLE_SCALA_VERSION }}-linux_amd64-*.jar | head -n 1) + if [ -z "$jar" ] || [ ! -f "$jar" ]; then Review Comment: Picking the first match via `ls ... | head -n 1` can select an unintended artifact if multiple matching jars exist (e.g., reruns, leftover files, or additional classifiers). Prefer a deterministic selection strategy (e.g., fail if there isn’t exactly one match, or select the newest by mtime) to make the workflow behavior predictable. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
