changes per comments.

core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala

@@ -67,7 +67,7 @@ private[spark] class ClusterTaskSetManager(
   val CPUS_PER_TASK = System.getProperty("spark.task.cpus", "1").toDouble
 
   // Maximum times a task is allowed to fail before failing the job
-  val MAX_TASK_FAILURES = System.getProperty("spark.task.max.fail", "4").toInt
+  val MAX_TASK_FAILURES = System.getProperty("spark.task.maxFailures", "4").toInt
 
   // Quantile of tasks at which to start speculation
   val SPECULATION_QUANTILE = System.getProperty("spark.speculation.quantile", "0.75").toDouble

docs/configuration.md

@@ -260,6 +260,14 @@ Apart from these, the following properties are also available, and may be useful
     applications). Note that any RDD that persists in memory for more than this duration will be cleared as well.
   </td>
 </tr>
+<tr>
+  <td>spark.task.maxFailures</td>
+  <td>4</td>
+  <td>
+    Number of individual task failures before giving up on the job.
+    Should greater or equal to 1. Number of allowed retries = this value - 1.
+  </td>
+</tr>
 
 </table>