test(spark-lineage): minor tweaks (#9717)

.github/workflows/spark-smoke-test.yml

@@ -42,8 +42,12 @@ jobs:
           cache: "pip"
       - name: Install dependencies
         run: ./metadata-ingestion/scripts/install_deps.sh
+      - name: Disk Check
+        run: df -h . && docker images
       - name: Remove images
         run: docker image prune -a -f || true
+      - name: Disk Check
+        run: df -h . && docker images
       - name: Smoke test
         run: |
           ./gradlew :metadata-integration:java:spark-lineage:integrationTest \
@@ -54,12 +58,24 @@ jobs:
              -x :datahub-web-react:yarnBuild \
              -x :datahub-web-react:distZip \
              -x :datahub-web-react:jar
+      - name: store logs
+        if: failure()
+        run: |
+          docker ps -a
+          docker logs datahub-gms >& gms-${{ matrix.test_strategy }}.log || true
+          docker logs datahub-actions >& actions-${{ matrix.test_strategy }}.log || true
+          docker logs broker >& broker-${{ matrix.test_strategy }}.log || true
+          docker logs mysql >& mysql-${{ matrix.test_strategy }}.log || true
+          docker logs elasticsearch >& elasticsearch-${{ matrix.test_strategy }}.log || true
+          docker logs datahub-frontend-react >& frontend-${{ matrix.test_strategy }}.log || true
       - name: Upload logs
         uses: actions/upload-artifact@v3
         if: failure()
         with:
           name: docker logs
-          path: "docker/build/container-logs/*.log"
+          path: |
+            "**/build/container-logs/*.log"
+            "*.log"
       - uses: actions/upload-artifact@v3
         if: always()
         with:

docker/build.gradle

@@ -8,15 +8,17 @@ import com.avast.gradle.dockercompose.tasks.ComposeDownForced
 apply from: "../gradle/versioning/versioning.gradle"
 
 ext {
-    quickstart_modules = [
+    backend_profile_modules = [
             ':docker:elasticsearch-setup',
             ':docker:mysql-setup',
             ':docker:kafka-setup',
             ':datahub-upgrade',
+            ':metadata-service:war',
+    ]
+    quickstart_modules = backend_profile_modules + [
             ':metadata-jobs:mce-consumer-job',
             ':metadata-jobs:mae-consumer-job',
-            ':metadata-service:war',
+            ':datahub-frontend'
-            ':datahub-frontend',
     ]
 
     debug_modules = quickstart_modules - [':metadata-jobs:mce-consumer-job',
@@ -90,9 +92,14 @@ dockerCompose {
         removeVolumes = false
     }
 
+    /**
+     * The smallest disk footprint required for Spark integration tests
+     *
+     * No frontend, mae, mce, or other services
+     */
     quickstartSlim {
         isRequiredBy(tasks.named('quickstartSlim'))
-        composeAdditionalArgs = ['--profile', 'quickstart-consumers']
+        composeAdditionalArgs = ['--profile', 'quickstart-backend']
 
         environment.put 'DATAHUB_VERSION', "v${version}"
         environment.put "DATAHUB_ACTIONS_IMAGE", "acryldata/datahub-ingestion"
@@ -132,7 +139,7 @@ tasks.getByName('quickstartComposeUp').dependsOn(
 tasks.getByName('quickstartPgComposeUp').dependsOn(
         pg_quickstart_modules.collect { it + ':dockerTag' })
 tasks.getByName('quickstartSlimComposeUp').dependsOn(
-        ([':docker:datahub-ingestion'] + quickstart_modules)
+        ([':docker:datahub-ingestion'] + backend_profile_modules)
                 .collect { it + ':dockerTag' })
 tasks.getByName('quickstartDebugComposeUp').dependsOn(
         debug_modules.collect { it + ':dockerTagDebug' }

metadata-integration/java/spark-lineage/src/test/java/datahub/spark/TestSparkJobsLineage.java

@@ -136,6 +136,7 @@ public class TestSparkJobsLineage {
         .respond(HttpResponse.response().withStatusCode(200));
   }
 
+  @BeforeClass
   public static void init() {
     mockServer = startClientAndServer(GMS_PORT);
     resetBaseExpectations();
@@ -219,8 +220,12 @@ public class TestSparkJobsLineage {
 
   @AfterClass
   public static void tearDown() throws Exception {
-    spark.stop();
+    if (spark != null) {
-    mockServer.stop();
+      spark.stop();
+    }
+    if (mockServer != null) {
+      mockServer.stop();
+    }
   }
 
   private static void check(List<DatasetLineage> expected, List<DatasetLineage> actual) {