[GitHub] [spark] jerryshao commented on a change in pull request #25309: [SPARK-28577][YARN]Ensure executorMemoryOverHead requested value not less than offHeapSize when offHeap enable

[GitBox <gi...@apache.org>]

jerryshao commented on a change in pull request #25309: [SPARK-28577][YARN]Ensure executorMemoryOverHead requested value not less than offHeapSize when offHeap enable
URL: https://github.com/apache/spark/pull/25309#discussion_r310108487
 
 

 ##########
 File path: resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnSparkHadoopUtil.scala
 ##########
 @@ -184,4 +184,29 @@ object YarnSparkHadoopUtil {
     ConverterUtils.toContainerId(containerIdString)
   }
 
+  /**
+   * If MEMORY_OFFHEAP_ENABLED is true, we should ensure executorOverheadMemory requested value
+   * is not less than MEMORY_OFFHEAP_SIZE, otherwise the memory resource requested for executor
+   * may be not enough.
+   */
+  def executorMemoryOverheadRequested(sparkConf: SparkConf): Int = {
+    val executorMemory = sparkConf.get(EXECUTOR_MEMORY).toInt
+    val overhead = sparkConf.get(EXECUTOR_MEMORY_OVERHEAD).getOrElse(
+      math.max((MEMORY_OVERHEAD_FACTOR * executorMemory).toInt, MEMORY_OVERHEAD_MIN)).toInt
+    val offHeap = if (sparkConf.get(MEMORY_OFFHEAP_ENABLED)) {
+      val size =
+        sparkConf.getSizeAsMb(MEMORY_OFFHEAP_SIZE.key, MEMORY_OFFHEAP_SIZE.defaultValueString)
+      require(size > 0,
+        s"${MEMORY_OFFHEAP_SIZE.key} must be > 0 when ${MEMORY_OFFHEAP_ENABLED.key} == true")
+      if (size > overhead) {
+        logWarning(s"The value of ${MEMORY_OFFHEAP_SIZE.key}(${size}MB) will be used as " +
+          s"executorMemoryOverhead to request resource to ensure that Executor has enough memory " +
+          s"to use. It is recommended that the configuration value of " +
+          s"${EXECUTOR_MEMORY_OVERHEAD.key} should be no less than ${MEMORY_OFFHEAP_SIZE.key} " +
+          s"when ${MEMORY_OFFHEAP_ENABLED.key} is true.")
+      }
+      size
+    } else 0
+    math.max(overhead, offHeap).toInt
 
 Review comment:
   I was wondering if it is better to change to `overhead = overhead + offHeap` if off-heap is enabled. Mainly because  off heap memory is not only used for tungsten related, but also for Netty and other native libraries. If we only guarantee `overhead > offHeap`, then it would somehow occupy the usage of Netty and others. Just my two cents :).

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org