我运行的YarnSpark工作，我有问题的容器大小。我以集群模式部署它。开始时，我在日志中看到下面一行。

INFO yarn.Client: Will allocate AM container, with 1408 MB memory including 384 MB overhead

偶尔，由于工作相当大，我会

INFO - ERROR yarn.Client: Application diagnostics message: Application application_1606984713739_1759 failed 2 times due to AM Container for appattempt_1606984713739_1759_000002 exited with  exitCode: -104    
INFO - Failing this attempt.Diagnostics: Container [pid=386064,containerID=container_e69_1606984713739_1759_02_000001] is running 5566464B beyond the 'PHYSICAL' memory limit. Current usage: 1.5 GB of 1.5 GB physical memory used; 3.3 GB of 3.1 GB virtual memory used. Killing container.

如何增加am尺寸？我的理解是，这需要记忆 spark.driver.memory ，或与之相近的东西 yarn.scheduler.minimum-allocation-mb （我将最小值设置为1GB，最大值设置为32GiB）。
给定作业的spark配置如下所示（我可以在spark ui环境中看到这一点，所以它确实已经设置好了）：

(spark.submit.deployMode,cluster)
...
(spark.executor.extraJavaOptions,-XX:InitiatingHeapOccupancyPercent=35)
(spark.executor.memory,20g)
(spark.executor.cores,5)
(spark.executor.instances,8)
(spark.driver.memoryOverhead,4g)
(spark.driver.memory,12g)
(spark.dynamicAllocation.minExecutors,4)
(spark.dynamicAllocation.maxExecutors,30)
(spark.dynamicAllocation.enabled,true)

是什么原因让我变得这么小？