[VL] Fix flaky columnar cache tests by clearing the cache serializer states before the suites are run

backends-velox/src/test/scala/org/apache/gluten/execution/ArrowCsvScanSuite.scala

@@ -21,6 +21,7 @@ import org.apache.gluten.datasource.ArrowCSVFileFormat
 
 import org.apache.spark.SparkConf
 import org.apache.spark.sql.execution.{ArrowFileSourceScanExec, BaseArrowScanExec, ColumnarToRowExec}
+import org.apache.spark.sql.execution.columnar.SparkCacheUtil
 import org.apache.spark.sql.types.{IntegerType, StringType, StructType}
 
 import org.scalatest.Ignore
@@ -83,6 +84,18 @@ class ArrowCsvScanWithTableCacheSuite extends ArrowCsvScanSuiteBase {
       .set(GlutenConfig.COLUMNAR_TABLE_CACHE_ENABLED.key, "true")
   }
 
+  override def beforeAll(): Unit = {
+    super.beforeAll()
+    // A common practice as well as in Spark tests, to clear the cache serializer
+    // in case it was already set as the default row-based serializer.
+    SparkCacheUtil.clearCacheSerializer()
+  }
+
+  override def afterAll(): Unit = {
+    SparkCacheUtil.clearCacheSerializer()
+    super.afterAll()
+  }
+
   /**
    * Test for GLUTEN-8453: https://github.com/apache/incubator-gluten/issues/8453. To make sure no
    * error is thrown when caching an Arrow Java query plan.

backends-velox/src/test/scala/org/apache/gluten/execution/VeloxColumnarCacheSuite.scala

@@ -23,7 +23,7 @@ import org.apache.spark.SparkConf
 import org.apache.spark.sql.Row
 import org.apache.spark.sql.execution.{ColumnarToRowExec, SparkPlan}
 import org.apache.spark.sql.execution.adaptive.AdaptiveSparkPlanHelper
-import org.apache.spark.sql.execution.columnar.InMemoryTableScanExec
+import org.apache.spark.sql.execution.columnar.{InMemoryTableScanExec, SparkCacheUtil}
 import org.apache.spark.sql.types.{LongType, Metadata, MetadataBuilder, StructType}
 import org.apache.spark.storage.StorageLevel
 
@@ -35,9 +35,17 @@ class VeloxColumnarCacheSuite extends VeloxWholeStageTransformerSuite with Adapt
 
   override def beforeAll(): Unit = {
     super.beforeAll()
+    // A common practice as well as in Spark tests, to clear the cache serializer
+    // in case it was already set as the default row-based serializer.
+    SparkCacheUtil.clearCacheSerializer()
     createTPCHNotNullTables()
   }
 
+  override protected def afterAll(): Unit = {
+    SparkCacheUtil.clearCacheSerializer()
+    super.afterAll()
+  }
+
   override protected def sparkConf: SparkConf = {
     super.sparkConf
       .set("spark.shuffle.manager", "org.apache.spark.shuffle.sort.ColumnarShuffleManager")

gluten-core/src/main/scala/org/apache/spark/sql/execution/columnar/SparkCacheUtil.scala

@@ -0,0 +1,23 @@
+/*
+ * 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.
+ */
+package org.apache.spark.sql.execution.columnar
+
+object SparkCacheUtil {
+  def clearCacheSerializer(): Unit = {
+    InMemoryRelation.clearSerializer()
+  }
+}

gluten-substrait/src/main/scala/org/apache/gluten/config/GlutenConfig.scala

@@ -963,7 +963,7 @@ object GlutenConfig {
       .createWithDefault(100000)
 
   val COLUMNAR_TABLE_CACHE_ENABLED =
-    buildConf("spark.gluten.sql.columnar.tableCache")
+    buildStaticConf("spark.gluten.sql.columnar.tableCache")
       .doc("Enable or disable columnar table cache.")
       .booleanConf
       .createWithDefault(false)