diff options
-rw-r--r-- | core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala | 2 | ||||
-rw-r--r-- | docs/configuration.md | 8 |
2 files changed, 9 insertions, 1 deletions
diff --git a/core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala b/core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala index bbf234febd..ffb5890ec2 100644 --- a/core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala +++ b/core/src/main/scala/spark/scheduler/cluster/ClusterTaskSetManager.scala @@ -85,7 +85,7 @@ private[spark] class ClusterTaskSetManager(sched: ClusterScheduler, val taskSet: 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 = 4 + 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 diff --git a/docs/configuration.md b/docs/configuration.md index 0bcd73ca99..99624a44aa 100644 --- a/docs/configuration.md +++ b/docs/configuration.md @@ -310,6 +310,14 @@ Apart from these, the following properties are also available, and may be useful Duration (milliseconds) of how long to batch new objects coming from network receivers. </td> </tr> +<tr> + <td>spark.task.maxFailures</td> + <td>4</td> + <td> + Number of individual task failures before giving up on the job. + Should be greater than or equal to 1. Number of allowed retries = this value - 1. + </td> +</tr> </table> |