ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[SPARK-11751] Doc describe error in the "Spark Streaming Programming Guide" page

Browse source 
In the **[Task Launching Overheads](http://spark.apache.org/docs/latest/streaming-programming-guide.html#task-launching-overheads)** section,
>Task Serialization: Using Kryo serialization for serializing tasks can reduce the task sizes, and therefore reduce the time taken to send them to the slaves.

as we known **Task Serialization** is configuration by **spark.closure.serializer** parameter, but currently only the Java serializer is supported. If we set **spark.closure.serializer** to **org.apache.spark.serializer.KryoSerializer**, then this will throw a exception.

Author: yangping.wu <wyphao.2007@163.com>

Closes #9734 from 397090770/397090770-patch-1.
This commit is contained in:
yangping.wu 2015-11-17 14:11:34 +00:00 committed by Sean Owen
parent fa13301ae4
commit 7276fa9aa9
1 changed files with 1 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
docs/streaming-programming-guide.md
View file

@ -2001,8 +2001,7 @@ If the number of tasks launched per second is high (say, 50 or more per second),
of sending out tasks to the slaves may be significant and will make it hard to achieve sub-second
latencies. The overhead can be reduced by the following changes:
* **Task Serialization**: Using Kryo serialization for serializing tasks can reduce the task
sizes, and therefore reduce the time taken to send them to the slaves.
* **Task Serialization**: Using Kryo serialization for serializing tasks can reduce the task sizes, and therefore reduce the time taken to send them to the slaves. This is controlled by the ```spark.closure.serializer``` property. However, at this time, Kryo serialization cannot be enabled for closure serialization. This may be resolved in a future release.
* **Execution mode**: Running Spark in Standalone mode or coarse-grained Mesos mode leads to
better task launch times than the fine-grained Mesos mode. Please refer to the