> 通过spark-submit会固定占用一占的资源,有什么办法,在任务不运作的时候将资源释放,让其它任务使用呢,yarn新版本默认已经支持了,我们使用的是HDP。
## 版本如下
![](https://upload-images.jianshu.io/upload_images/9028759-35c1bf0606261dc5.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
## 配置
1. HDP里面已经默认支持spark动态资源释配置
2. 代码配置
```
val sparkConf = new SparkConf()
.set("spark.shuffle.service.enabled", "true")
.set("spark.dynamicAllocation.enabled", "true")
.set("spark.dynamicAllocation.minExecutors", "1") //最少占用1个Executor
.set("spark.dynamicAllocation.initialExecutors", "1") //默认初始化一个Executor
.set("spark.dynamicAllocation.maxExecutors", "6") //最多占用6个Executor
.set("spark.dynamicAllocation.executorIdleTimeout", "60") //executor闲置时间
.set("spark.dynamicAllocation.cachedExecutorIdleTimeout", "60") //cache闲置时间
.set("spark.executor.cores", "3")//使用的vcore
// .setMaster("local[12]")
.setAppName("Spark DynamicRelease")
val spark: SparkSession = SparkSession
.builder
.config(sparkConf)
.getOrCreate()
```
## 注意事项
如果spark计算当中使用了rdd.cache,不加下面的配置,动态资源不会释放
```
.set("spark.dynamicAllocation.cachedExecutorIdleTimeout", "60")
```
---
![](https://upload-images.jianshu.io/upload_images/9028759-07315bb8dadcd082.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)