[SPARK-45265][SQL] Support Hive 4.0 metastore

docs/sql-data-sources-hive-tables.md

@@ -130,7 +130,7 @@ The following options can be used to configure the version of Hive that is used
     <td><code>2.3.10</code></td>
     <td>
       Version of the Hive metastore. Available
-      options are <code>2.0.0</code> through <code>2.3.10</code> and <code>3.0.0</code> through <code>3.1.3</code>.
+      options are <code>2.0.0</code> through <code>2.3.10</code>, <code>3.0.0</code> through <code>3.1.3</code>, and <code>4.0.0</code> through <code>4.0.1</code>.
     </td>
     <td>1.4.0</td>
   </tr>

project/SparkBuild.scala

@@ -1664,7 +1664,7 @@ object TestSettings {
         "--add-opens=java.base/sun.util.calendar=ALL-UNNAMED",
         "-Djdk.reflect.useDirectMethodHandle=false",
         "-Dio.netty.tryReflectionSetAccessible=true").mkString(" ")
-      s"-Xmx$heapSize -Xss4m -XX:MaxMetaspaceSize=$metaspaceSize -XX:ReservedCodeCacheSize=128m -Dfile.encoding=UTF-8 $extraTestJavaArgs"
+      s"-Xmx$heapSize -Xss64m -XX:MaxMetaspaceSize=$metaspaceSize -XX:ReservedCodeCacheSize=128m -Dfile.encoding=UTF-8 $extraTestJavaArgs"
         .split(" ").toSeq
     },
     javaOptions ++= {

sql/hive-thriftserver/src/main/scala/org/apache/spark/sql/hive/thriftserver/SparkSQLCLIDriver.scala

@@ -458,27 +458,28 @@ private[hive] class SparkSQLCLIDriver extends CliDriver with Logging {
           if (sessionState.getIsVerbose) {
             out.println(cmd)
           }
-          val rc = driver.run(cmd)
-          val endTimeNs = System.nanoTime()
-          val timeTaken: Double = TimeUnit.NANOSECONDS.toMillis(endTimeNs - startTimeNs) / 1000.0
-
-          ret = rc.getResponseCode
-          if (ret != 0) {
-            val format = SparkSQLEnv.sparkSession.sessionState.conf.errorMessageFormat
-            val e = rc.getException
-            val msg = e match {
-              case st: SparkThrowable with Throwable => SparkThrowableHelper.getMessage(st, format)
-              case _ => e.getMessage
-            }
-            err.println(msg)
-            if (format == ErrorMessageFormat.PRETTY &&
+          try {
+            driver.run(cmd)
+          } catch {
+            case t: Throwable =>
+              ret = 1
+              val format = SparkSQLEnv.sparkSession.sessionState.conf.errorMessageFormat
+              val msg = t match {
+                case st: SparkThrowable with Throwable =>
+                  SparkThrowableHelper.getMessage(st, format)
+                case _ => t.getMessage
+              }
+              err.println(msg)
+              if (format == ErrorMessageFormat.PRETTY &&
                 !sessionState.getIsSilent &&
-                (!e.isInstanceOf[AnalysisException] || e.getCause != null)) {
-              e.printStackTrace(err)
-            }
-            driver.close()
-            return ret
+                (!t.isInstanceOf[AnalysisException] || t.getCause != null)) {
+                t.printStackTrace(err)
+              }
+              driver.close()
+              return ret
           }
+          val endTimeNs = System.nanoTime()
+          val timeTaken: Double = TimeUnit.NANOSECONDS.toMillis(endTimeNs - startTimeNs) / 1000.0
 
           val res = new JArrayList[String]()
 

sql/hive-thriftserver/src/main/scala/org/apache/spark/sql/hive/thriftserver/SparkSQLDriver.scala

@@ -31,7 +31,7 @@ import org.apache.spark.internal.{Logging, MDC}
 import org.apache.spark.internal.LogKeys.COMMAND
 import org.apache.spark.sql.SparkSession
 import org.apache.spark.sql.catalyst.plans.logical.CommandResult
-import org.apache.spark.sql.execution.{QueryExecution, SQLExecution}
+import org.apache.spark.sql.execution.{QueryExecution, QueryExecutionException, SQLExecution}
 import org.apache.spark.sql.execution.HiveResult.hiveResultString
 import org.apache.spark.sql.internal.{SQLConf, VariableSubstitution}
 
@@ -82,10 +82,10 @@ private[hive] class SparkSQLDriver(val sparkSession: SparkSession = SparkSQLEnv.
     } catch {
         case st: SparkThrowable =>
           logDebug(s"Failed in [$command]", st)
-          new CommandProcessorResponse(1, ExceptionUtils.getStackTrace(st), st.getSqlState, st)
+          throw st
         case cause: Throwable =>
           logError(log"Failed in [${MDC(COMMAND, command)}]", cause)
-          new CommandProcessorResponse(1, ExceptionUtils.getStackTrace(cause), null, cause)
+          throw new QueryExecutionException(ExceptionUtils.getStackTrace(cause))
     }
   }
 

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveExternalCatalog.scala

@@ -1030,7 +1030,7 @@ private[spark] class HiveExternalCatalog(conf: SparkConf, hadoopConf: Configurat
     }
     val metaStoreParts = partsWithLocation
       .map(p => p.copy(spec = toMetaStorePartitionSpec(p.spec)))
-    client.createPartitions(db, table, metaStoreParts, ignoreIfExists)
+    client.createPartitions(tableMeta, metaStoreParts, ignoreIfExists)
   }
 
   override def dropPartitions(

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveUtils.scala

@@ -74,8 +74,9 @@ private[spark] object HiveUtils extends Logging {
 
   val HIVE_METASTORE_VERSION = buildStaticConf("spark.sql.hive.metastore.version")
     .doc("Version of the Hive metastore. Available options are " +
-        "<code>2.0.0</code> through <code>2.3.10</code> and " +
-        "<code>3.0.0</code> through <code>3.1.3</code>.")
+      "<code>2.0.0</code> through <code>2.3.10</code>, " +
+      "<code>3.0.0</code> through <code>3.1.3</code> and " +
+      "<code>4.0.0</code> through <code>4.0.1</code>.")
     .version("1.4.0")
     .stringConf
     .checkValue(isCompatibleHiveVersion, "Unsupported Hive Metastore version")

sql/hive/src/main/scala/org/apache/spark/sql/hive/client/HiveClient.scala

@@ -164,8 +164,7 @@ private[hive] trait HiveClient {
    * Create one or many partitions in the given table.
    */
   def createPartitions(
-      db: String,
-      table: String,
+      table: CatalogTable,
       parts: Seq[CatalogTablePartition],
       ignoreIfExists: Boolean): Unit
 

sql/hive/src/main/scala/org/apache/spark/sql/hive/client/HiveClientImpl.scala

@@ -17,7 +17,7 @@
 
 package org.apache.spark.sql.hive.client
 
-import java.io.PrintStream
+import java.io.{OutputStream, PrintStream}
 import java.lang.{Iterable => JIterable}
 import java.lang.reflect.InvocationTargetException
 import java.nio.charset.StandardCharsets.UTF_8
@@ -28,6 +28,7 @@ import scala.collection.mutable
 import scala.collection.mutable.ArrayBuffer
 import scala.jdk.CollectionConverters._
 
+import org.apache.commons.lang3.exception.ExceptionUtils
 import org.apache.hadoop.conf.Configuration
 import org.apache.hadoop.fs.Path
 import org.apache.hadoop.hive.common.StatsSetupConst
@@ -44,7 +45,7 @@ import org.apache.hadoop.hive.serde2.MetadataTypedColumnsetSerDe
 import org.apache.hadoop.hive.serde2.`lazy`.LazySimpleSerDe
 import org.apache.hadoop.security.UserGroupInformation
 
-import org.apache.spark.{SparkConf, SparkException}
+import org.apache.spark.{SparkConf, SparkException, SparkThrowable}
 import org.apache.spark.deploy.SparkHadoopUtil.SOURCE_SPARK
 import org.apache.spark.internal.{Logging, LogKeys, MDC}
 import org.apache.spark.internal.LogKeys._
@@ -121,6 +122,7 @@ private[hive] class HiveClientImpl(
     case hive.v2_3 => new Shim_v2_3()
     case hive.v3_0 => new Shim_v3_0()
     case hive.v3_1 => new Shim_v3_1()
+    case hive.v4_0 => new Shim_v4_0()
   }
 
   // Create an internal session state for this HiveClientImpl.
@@ -177,8 +179,10 @@ private[hive] class HiveClientImpl(
     // got changed. We reset it to clientLoader.ClassLoader here.
     state.getConf.setClassLoader(clientLoader.classLoader)
     shim.setCurrentSessionState(state)
-    state.out = new PrintStream(outputBuffer, true, UTF_8.name())
-    state.err = new PrintStream(outputBuffer, true, UTF_8.name())
+    val clz = state.getClass.getField("out").getType.asInstanceOf[Class[_ <: PrintStream]]
+    val ctor = clz.getConstructor(classOf[OutputStream], classOf[Boolean], classOf[String])
+    state.getClass.getField("out").set(state, ctor.newInstance(outputBuffer, true, UTF_8.name()))
+    state.getClass.getField("err").set(state, ctor.newInstance(outputBuffer, true, UTF_8.name()))
     state
   }
 
@@ -307,15 +311,27 @@ private[hive] class HiveClientImpl(
   }
 
   def setOut(stream: PrintStream): Unit = withHiveState {
-    state.out = stream
+    val ctor = state.getClass.getField("out")
+      .getType
+      .asInstanceOf[Class[_ <: PrintStream]]
+      .getConstructor(classOf[OutputStream])
+    state.getClass.getField("out").set(state, ctor.newInstance(stream))
   }
 
   def setInfo(stream: PrintStream): Unit = withHiveState {
-    state.info = stream
+    val ctor = state.getClass.getField("info")
+      .getType
+      .asInstanceOf[Class[_ <: PrintStream]]
+      .getConstructor(classOf[OutputStream])
+    state.getClass.getField("info").set(state, ctor.newInstance(stream))
   }
 
   def setError(stream: PrintStream): Unit = withHiveState {
-    state.err = stream
+    val ctor = state.getClass.getField("err")
+      .getType
+      .asInstanceOf[Class[_ <: PrintStream]]
+      .getConstructor(classOf[OutputStream])
+    state.getClass.getField("err").set(state, ctor.newInstance(stream))
   }
 
   private def setCurrentDatabaseRaw(db: String): Unit = {
@@ -629,21 +645,22 @@ private[hive] class HiveClientImpl(
   }
 
   override def createPartitions(
-      db: String,
-      table: String,
+      table: CatalogTable,
       parts: Seq[CatalogTablePartition],
       ignoreIfExists: Boolean): Unit = withHiveState {
     def replaceExistException(e: Throwable): Unit = e match {
       case _: HiveException if e.getCause.isInstanceOf[AlreadyExistsException] =>
-        val hiveTable = client.getTable(db, table)
+        val db = table.identifier.database.getOrElse(state.getCurrentDatabase)
+        val tableName = table.identifier.table
+        val hiveTable = client.getTable(db, tableName)
         val existingParts = parts.filter { p =>
           shim.getPartitions(client, hiveTable, p.spec.asJava).nonEmpty
         }
-        throw new PartitionsAlreadyExistException(db, table, existingParts.map(_.spec))
+        throw new PartitionsAlreadyExistException(db, tableName, existingParts.map(_.spec))
       case _ => throw e
     }
     try {
-      shim.createPartitions(client, db, table, parts, ignoreIfExists)
+      shim.createPartitions(client, toHiveTable(table), parts, ignoreIfExists)
     } catch {
       case e: InvocationTargetException => replaceExistException(e.getCause)
       case e: Throwable => replaceExistException(e)
@@ -861,11 +878,22 @@ private[hive] class HiveClientImpl(
       // Since HIVE-18238(Hive 3.0.0), the Driver.close function's return type changed
       // and the CommandProcessorFactory.clean function removed.
       driver.getClass.getMethod("close").invoke(driver)
-      if (version != hive.v3_0 && version != hive.v3_1) {
+      if (version != hive.v3_0 && version != hive.v3_1 && version != hive.v4_0) {
         CommandProcessorFactory.clean(conf)
       }
     }
 
+    def getResponseCode(response: CommandProcessorResponse): Int = {
+      if (version < hive.v4_0) {
+        response.getResponseCode
+      } else {
+        // Since Hive 4.0, response code is removed from CommandProcessorResponse.
+        // Here we simply return 0 for the positive cases as for error cases it will
+        // throw exceptions early.
+        0
+      }
+    }
+
     // Hive query needs to start SessionState.
     SessionState.start(state)
     logDebug(s"Running hiveql '$cmd'")
@@ -878,30 +906,44 @@ private[hive] class HiveClientImpl(
       val proc = shim.getCommandProcessor(tokens(0), conf)
       proc match {
         case driver: Driver =>
-          val response: CommandProcessorResponse = driver.run(cmd)
-          // Throw an exception if there is an error in query processing.
-          if (response.getResponseCode != 0) {
+          try {
+            val response: CommandProcessorResponse = driver.run(cmd)
+            if (getResponseCode(response) != 0) {
+              // Throw an exception if there is an error in query processing.
+              // This works for hive 3.x and earlier versions.
+              throw new QueryExecutionException(response.getErrorMessage)
+            }
+            driver.setMaxRows(maxRows)
+            val results = shim.getDriverResults(driver)
+            results
+          } catch {
+            case e @ (_: QueryExecutionException | _: SparkThrowable) =>
+              throw e
+            case e: Exception =>
+              // Wrap the original hive error with QueryExecutionException and throw it
+              // if there is an error in query processing.
+              // This works for hive 4.x and later versions.
+              throw new QueryExecutionException(ExceptionUtils.getStackTrace(e))
+          } finally {
             closeDriver(driver)
-            throw new QueryExecutionException(response.getErrorMessage)
           }
-          driver.setMaxRows(maxRows)
-
-          val results = shim.getDriverResults(driver)
-          closeDriver(driver)
-          results
 
         case _ =>
-          if (state.out != null) {
+          val out = state.getClass.getField("out").get(state)
+          if (out != null) {
             // scalastyle:off println
-            state.out.println(tokens(0) + " " + cmd_1)
+            out.asInstanceOf[PrintStream].println(tokens(0) + " " + cmd_1)
             // scalastyle:on println
           }
           val response: CommandProcessorResponse = proc.run(cmd_1)
-          // Throw an exception if there is an error in query processing.
-          if (response.getResponseCode != 0) {
+          val responseCode = getResponseCode(response)
+          if (responseCode != 0) {
+            // Throw an exception if there is an error in query processing.
+            // This works for hive 3.x and earlier versions. For 4.x and later versions,
+            // It will go to the catch block directly.
             throw new QueryExecutionException(response.getErrorMessage)
           }
-          Seq(response.getResponseCode.toString)
+          Seq(responseCode.toString)
       }
     } catch {
       case e: Exception =>
@@ -971,7 +1013,7 @@ private[hive] class HiveClientImpl(
       partSpec,
       replace,
       numDP,
-      listBucketingEnabled = hiveTable.isStoredAsSubDirectories)
+      hiveTable)
   }
 
   override def createFunction(db: String, func: CatalogFunction): Unit = withHiveState {