refactor: update kuttl tests

Author: razvan

tests/templates/kuttl/resources/10-assert.yaml.j2

@@ -9,7 +9,7 @@ apiVersion: v1
 kind: Pod
 metadata:
   labels:
-    app.kubernetes.io/instance: resources-crd
+    # app.kubernetes.io/instance: resources-crd
     spark-role: driver
 spec:
   containers:

tests/templates/kuttl/resources/10-deploy-spark-app.yaml.j2

@@ -20,7 +20,7 @@ spec:
   mainApplicationFile: "local:///stackable/spark/examples/jars/spark-examples.jar"
   sparkConf:
     spark.kubernetes.submission.waitAppCompletion: "false"
-    spark.kubernetes.driver.pod.name: "resources-crd-driver"
+    # spark.kubernetes.driver.pod.name: "resources-crd-driver"
     spark.kubernetes.executor.podNamePrefix: "resources-crd"
   job:
     config:

tests/templates/kuttl/resources/12-assert.yaml.j2

@@ -9,7 +9,6 @@ apiVersion: v1
 kind: Pod
 metadata:
   labels:
-    app.kubernetes.io/instance: resources-sparkconf
     spark-role: driver
 spec:
   containers:
@@ -19,11 +18,11 @@ spec:
     - name: spark
       resources:
         limits:
-          cpu: "1"
-          memory: "1433Mi"
+          cpu: "2"
+          memory: "2Gi"
         requests:
-          cpu: "1"
-          memory: "1433Mi"
+          cpu: "500m"
+          memory: "2Gi"
 ---
 apiVersion: v1
 kind: Pod

tests/templates/kuttl/resources/12-deploy-spark-app.yaml.j2

@@ -20,13 +20,15 @@ spec:
   mainApplicationFile: "local:///stackable/spark/examples/jars/spark-examples.jar"
   sparkConf:
     spark.kubernetes.submission.waitAppCompletion: "false"
-    spark.kubernetes.driver.pod.name: "resources-sparkconf-driver"
+    # NOTE: Setting the driver pod name is not possible anymore starting with v1alpha2
+    # spark.kubernetes.driver.pod.name: "resources-sparkconf-driver"
     spark.kubernetes.executor.podNamePrefix: "resources-sparkconf"
-    spark.kubernetes.driver.request.cores: "1"
-    spark.kubernetes.driver.limit.cores: "1"
-    spark.driver.cores: "1"
-    spark.driver.memory: "1g"
-    spark.driver.memoryOverheadFactor: "0.4"
+    # NOTE: Driver pod resources are NOT set via sparkConf. The operator runs the driver as a
+    # Kubernetes Job with spark-submit in client mode, so Spark never creates the driver pod and
+    # the spark.kubernetes.driver.{request,limit}.cores / spark.driver.memory properties are
+    # inert. The driver pod's resources come solely from spec.driver.config.resources below.
+    # Executor resources, on the other hand, ARE driven by sparkConf because the driver still
+    # asks Spark to create the executor pods.
     spark.kubernetes.executor.request.cores: "1"
     spark.kubernetes.executor.limit.cores: "2"
     spark.executor.cores: "2"
@@ -35,6 +37,14 @@ spec:
     spark.executor.instances: "1"
   driver:
     config:
+      # Values intentionally differ from the operator defaults (250m/1, 1Gi) to prove that
+      # user-provided role config takes priority. min -> request, max -> limit.
+      resources:
+        cpu:
+          min: "500m"
+          max: "2"
+        memory:
+          limit: "2Gi"
       logging:
         enableVectorAgent: {{ lookup('env', 'VECTOR_AGGREGATOR') | length > 0 }}
   executor:

tests/templates/kuttl/smoke/50-deploy-spark-app.yaml.j2

@@ -20,6 +20,12 @@ spec:
   # spark-submit loads the application jar to s3a://my-bucket and thus makes it
   # available to the driver and executors.
   mainApplicationFile: "/stackable/spark/examples/jars/spark-examples.jar"
+  # The argument is the number of "slices" (partitions). SparkPi runs one task
+  # per slice sequentially on the single executor, so a high value stretches the
+  # runtime to ~1-2 minutes. This gives the assert in 50-assert.yaml enough time
+  # to exec into the running driver pod and for containerdebug to start.
+  args:
+    - "15000"
   sparkConf:
     spark.kubernetes.file.upload.path: "s3a://my-bucket"
   s3connection:

tests/templates/kuttl/spark-examples/10-assert.yaml

@@ -4,7 +4,7 @@ kind: TestAssert
 timeout: 900
 ---
 # The Job starting the whole process
-apiVersion: spark.stackable.tech/v1alpha1
+apiVersion: spark.stackable.tech/v1alpha2
 kind: SparkApplication
 metadata:
   name: spark-examples

tests/templates/kuttl/spark-examples/10-deploy-spark-app.yaml.j2

@@ -34,6 +34,17 @@ spec:
     config:
       logging:
         enableVectorAgent: {{ lookup('env', 'VECTOR_AGGREGATOR') | length > 0 }}
+      # TODO: This is copied from the spec.job.config above because that is ignored in v1alpha2
+      affinity:
+        nodeAffinity:
+          preferredDuringSchedulingIgnoredDuringExecution:
+          - weight: 1
+            preference:
+              matchExpressions:
+              - key: disktype
+                operator: In
+                values:
+                - ssd
   executor:
     replicas: 1
     config: