ci: add Spark 4.0 / JDK 21 profile (apache#4060)

Author: james-willis

.github/workflows/pr_build_linux.yml

@@ -81,7 +81,7 @@ jobs:
     container:
       image: amd64/rust
       env:
-        JAVA_TOOL_OPTIONS: ${{ matrix.profile.java_version == '17' && '--add-exports=java.base/sun.nio.ch=ALL-UNNAMED --add-exports=java.base/sun.util.calendar=ALL-UNNAMED --add-opens=java.base/java.nio=ALL-UNNAMED --add-opens=java.base/java.lang=ALL-UNNAMED' || '' }}
+        JAVA_TOOL_OPTIONS: ${{ (matrix.profile.java_version == '17' || matrix.profile.java_version == '21') && '--add-exports=java.base/sun.nio.ch=ALL-UNNAMED --add-exports=java.base/sun.util.calendar=ALL-UNNAMED --add-opens=java.base/java.nio=ALL-UNNAMED --add-opens=java.base/java.lang=ALL-UNNAMED' || '' }}
     strategy:
       matrix:
         profile:
@@ -94,6 +94,9 @@ jobs:
           - name: "Spark 4.0, JDK 17"
             java_version: "17"
             maven_opts: "-Pspark-4.0"
+          - name: "Spark 4.0, JDK 21"
+            java_version: "21"
+            maven_opts: "-Pspark-4.0"
           # Spark 4.1 is intentionally absent: the lint job invokes -Psemanticdb,
           # but semanticdb-scalac_2.13.17 is not yet published, so we cannot
           # currently run scalafix against the spark-4.1 profile.
@@ -293,6 +296,11 @@ jobs:
             maven_opts: "-Pspark-4.0"
             scan_impl: "auto"
 
+          - name: "Spark 4.0, JDK 21"
+            java_version: "21"
+            maven_opts: "-Pspark-4.0"
+            scan_impl: "auto"
+
           - name: "Spark 4.1, JDK 17"
             java_version: "17"
             maven_opts: "-Pspark-4.1"
@@ -388,7 +396,7 @@ jobs:
     container:
       image: amd64/rust
       env:
-        JAVA_TOOL_OPTIONS: ${{ matrix.profile.java_version == '17' && '--add-exports=java.base/sun.nio.ch=ALL-UNNAMED --add-exports=java.base/sun.util.calendar=ALL-UNNAMED --add-opens=java.base/java.nio=ALL-UNNAMED --add-opens=java.base/java.lang=ALL-UNNAMED' || '' }}
+        JAVA_TOOL_OPTIONS: ${{ (matrix.profile.java_version == '17' || matrix.profile.java_version == '21') && '--add-exports=java.base/sun.nio.ch=ALL-UNNAMED --add-exports=java.base/sun.util.calendar=ALL-UNNAMED --add-opens=java.base/java.nio=ALL-UNNAMED --add-opens=java.base/java.lang=ALL-UNNAMED' || '' }}
 
     steps:
       - uses: runs-on/action@742bf56072eb4845a0f94b3394673e4903c90ff0  # v2.1.0

.github/workflows/spark_sql_test.yml

@@ -141,6 +141,7 @@ jobs:
           - {spark-short: '3.4', spark-full: '3.4.3', java: 11, scan-impl: 'auto'}
           - {spark-short: '3.5', spark-full: '3.5.8', java: 11, scan-impl: 'auto'}
           - {spark-short: '4.0', spark-full: '4.0.2', java: 17, scan-impl: 'auto'}
+          - {spark-short: '4.0', spark-full: '4.0.2', java: 21, scan-impl: 'auto'}
       fail-fast: false
     name: spark-sql-${{ matrix.config.scan-impl }}-${{ matrix.module.name }}/spark-${{ matrix.config.spark-full }}
     runs-on: ${{ matrix.os }}
@@ -175,6 +176,14 @@ jobs:
           fi
         env:
           LC_ALL: "C.UTF-8"
+          # Mirror Spark's own JDK 21 / 25 CI workaround. apache/spark's
+          # build_java21.yml and build_java25.yml set this same env var to
+          # process-isolate the V1/V2 Parquet and Orc source suites because
+          # they exhibit cross-suite resource interactions (file-stream and
+          # thread leaks) under the newer JDKs. project/SparkBuild.scala
+          # reads DEDICATED_JVM_SBT_TESTS and forks a separate JVM per
+          # listed suite. Empty value is a safe no-op.
+          DEDICATED_JVM_SBT_TESTS: ${{ matrix.config.spark-short == '4.0' && 'org.apache.spark.sql.execution.datasources.parquet.ParquetFileFormatV1Suite,org.apache.spark.sql.execution.datasources.parquet.ParquetFileFormatV2Suite,org.apache.spark.sql.execution.datasources.orc.OrcSourceV1Suite,org.apache.spark.sql.execution.datasources.orc.OrcSourceV2Suite' || '' }}
       - name: Upload fallback log
         if: ${{ github.event.inputs.collect-fallback-logs == 'true' }}
         uses: actions/upload-artifact@v7

docs/source/user-guide/latest/installation.md

@@ -43,7 +43,7 @@ Other versions may work well enough for development and evaluation purposes.
 | ------------- | ------------ | ------------- | ----------------- | --------------------- |
 | 3.4.3         | 11/17        | 2.12/2.13     | Yes               | Yes                   |
 | 3.5.8         | 11/17        | 2.12/2.13     | Yes               | Yes                   |
-| 4.0.2         | 17           | 2.13          | Yes               | Yes                   |
+| 4.0.2         | 17/21        | 2.13          | Yes               | Yes                   |
 
 Note that we do not test the full matrix of supported Java and Scala versions in CI for every Spark version.